agora inbox for [email protected]
help / color / mirror / Atom feed[PATCH v9 2/8] Implement CLUSTER of partitioned table..
152+ messages / 7 participants
[nested] [flat]
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 1/5] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 167 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/nodes/parsenodes.h | 5 +-
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 53 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index b9450e7366..0476cfff72 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -172,6 +172,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..391e018bbd 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +74,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -113,7 +118,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -124,14 +129,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +166,32 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +201,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +224,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -328,9 +330,10 @@ cluster_rel(Oid tableOid, Oid indexOid, int options)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -374,8 +377,13 @@ cluster_rel(Oid tableOid, Oid indexOid, int options)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -391,6 +399,14 @@ cluster_rel(Oid tableOid, Oid indexOid, int options)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -459,6 +475,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -483,12 +502,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -560,10 +573,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1557,3 +1566,75 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 5238a960f7..07ef7fc1b7 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 53de31f3b1..087b1c7af6 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -3199,8 +3199,9 @@ typedef struct AlterSystemStmt
*/
typedef enum ClusterOption
{
- CLUOPT_RECHECK = 1 << 0, /* recheck relation state */
- CLUOPT_VERBOSE = 1 << 1 /* print progress info */
+ CLUOPT_VERBOSE = 1 << 0, /* print progress info */
+ CLUOPT_RECHECK = 1 << 1, /* recheck relation state */
+ CLUOPT_RECHECK_ISCLUSTERED = 1 << 2, /* recheck relation state for indisclustered */
} ClusterOption;
typedef struct ClusterStmt
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--Sr1nOIr3CvdE5hEN
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v4-0002-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v11] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
For now, partitioned indexes cannot be marked clustered, so clustering requires
specification of a partitioned index on the partitioned table.
VACUUM (including vacuum full) has recursed into partitione tables since
partitioning were introduced in v10 (3c3bb9933). See expand_vacuum_rel().
See also a556549d7 and 19de0ab23.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 170 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 46 ++++++-
src/test/regress/sql/cluster.sql | 22 +++-
6 files changed, 197 insertions(+), 49 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 86f5fdc469..b3463ae5c4 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -196,6 +196,12 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index (which must be specified).
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 9d22f648a8..c5923c5217 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -131,6 +136,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
Relation rel;
/* Find, lock, and check permissions on the table */
+ /* Obtain AEL now to avoid lock-upgrade hazard in the single-transaction case */
tableOid = RangeVarGetRelidExtended(stmt->relation,
AccessExclusiveLock,
0,
@@ -146,14 +152,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -188,11 +186,45 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
stmt->indexname, stmt->relation->relname)));
}
- /* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* close relation, keep lock till commit */
+ table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+
+ /*
+ * For now, partitioned indexes are not actually marked clustered.
+ */
+ check_index_is_clusterable(rel, indexOid, true, AccessShareLock);
+
+ /* close relation, releasing lock on parent table */
+ table_close(rel, AccessExclusiveLock);
+
+ /* Do the job. */
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +234,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +256,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -327,10 +337,11 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/*
* Silently skip a temp table for a remote session. Only doing this
* check in the "recheck" case is appropriate (which currently means
- * somebody is executing a database-wide CLUSTER), because there is
- * another check in cluster() which will stop any attempt to cluster
- * remote temp tables by name. There is another check in cluster_rel
- * which is redundant, but we leave it for extra safety.
+ * somebody is executing a database-wide CLUSTER or on a partitioned
+ * table), because there is another check in cluster() which will stop
+ * any attempt to cluster remote temp tables by name. There is another
+ * check in cluster_rel which is redundant, but we leave it for extra
+ * safety.
*/
if (RELATION_IS_OTHER_TEMP(OldHeap))
{
@@ -352,9 +363,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -415,6 +427,9 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ // Should silently skip this rather than assert ?
+ Assert(RELKIND_HAS_STORAGE(OldHeap->rd_rel->relkind));
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -585,8 +600,8 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
if (OidIsValid(indexOid))
+ /* Mark the correct index as clustered */
mark_index_clustered(OldHeap, indexOid, true);
/* Remember info about rel before closing OldHeap */
@@ -1604,3 +1619,74 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and their associated indexes, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(indexrelid)))
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 5cd5838668..c377511278 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -599,6 +599,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index b64d3bc204..d400b6e937 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index e46a66952f..7461e1c090 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -444,14 +444,52 @@ DROP TABLE clustertest;
CREATE TABLE clustertest (f1 int PRIMARY KEY);
CLUSTER clustertest USING clustertest_pkey;
CLUSTER clustertest;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+CLUSTER clstrpart USING clstrpart_idx;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Partitioned indexes aren't and can't be marked un/clustered:
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a)
+Number of partitions: 3 (Use \d+ to list them.)
+
+CLUSTER clstrpart;
+ERROR: there is no previously clustered index for table "clstrpart"
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+ERROR: cannot mark index clustered in partitioned table
ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
ERROR: cannot mark index clustered in partitioned table
-CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index aee9cf83e0..f31f8ef2b2 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -202,12 +202,28 @@ CREATE TABLE clustertest (f1 int PRIMARY KEY);
CLUSTER clustertest USING clustertest_pkey;
CLUSTER clustertest;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Partitioned indexes aren't and can't be marked un/clustered:
+\d clstrpart
+CLUSTER clstrpart;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--DwoPkXS38qd3dnhB--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 230 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/nodes/parsenodes.h | 5 +-
src/test/regress/expected/cluster.out | 70 +++++++-
src/test/regress/sql/cluster.sql | 27 ++-
6 files changed, 268 insertions(+), 71 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 4da60d8d56..2df8aaabc4 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -172,6 +172,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 0d647e912c..d2e3ebc8af 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -74,7 +76,11 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool *pSwapToastByContent,
TransactionId *pFreezeXid,
MultiXactId *pCutoffMulti);
+static void set_indisclustered(Oid indexOid, bool isclustered, Relation pg_index);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, bool isTopLevel, int options);
/*---------------------------------------------------------------------------
@@ -116,7 +122,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -127,14 +133,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -172,8 +170,32 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options, isTopLevel);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options, isTopLevel);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, isTopLevel, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -183,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -207,26 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK,
- isTopLevel);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, isTopLevel, stmt->options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -332,9 +334,10 @@ cluster_rel(Oid tableOid, Oid indexOid, int options, bool isTopLevel)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -378,8 +381,13 @@ cluster_rel(Oid tableOid, Oid indexOid, int options, bool isTopLevel)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -395,6 +403,14 @@ cluster_rel(Oid tableOid, Oid indexOid, int options, bool isTopLevel)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -463,6 +479,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -474,6 +493,32 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
index_close(OldIndex, NoLock);
}
+/*
+ * Helper for mark_index_clustered
+ * Mark a single index as clustered or not.
+ * pg_index is passed by caller to avoid repeatedly re-opening it.
+ */
+static void
+set_indisclustered(Oid indexOid, bool isclustered, Relation pg_index)
+{
+ HeapTuple indexTuple;
+ Form_pg_index indexForm;
+
+ indexTuple = SearchSysCacheCopy1(INDEXRELID,
+ ObjectIdGetDatum(indexOid));
+ if (!HeapTupleIsValid(indexTuple))
+ elog(ERROR, "cache lookup failed for index %u", indexOid);
+ indexForm = (Form_pg_index) GETSTRUCT(indexTuple);
+
+ /* this was checked earlier, but let's be real sure */
+ if (isclustered && !indexForm->indisvalid)
+ elog(ERROR, "cannot cluster on invalid index %u", indexOid);
+
+ indexForm->indisclustered = isclustered;
+ CatalogTupleUpdate(pg_index, &indexTuple->t_self, indexTuple);
+ heap_freetuple(indexTuple);
+}
+
/*
* mark_index_clustered: mark the specified index as the one clustered on
*
@@ -482,17 +527,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
void
mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
{
- HeapTuple indexTuple;
- Form_pg_index indexForm;
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -511,34 +548,32 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
{
Oid thisIndexOid = lfirst_oid(index);
- indexTuple = SearchSysCacheCopy1(INDEXRELID,
- ObjectIdGetDatum(thisIndexOid));
- if (!HeapTupleIsValid(indexTuple))
- elog(ERROR, "cache lookup failed for index %u", thisIndexOid);
- indexForm = (Form_pg_index) GETSTRUCT(indexTuple);
-
/*
* Unset the bit if set. We know it's wrong because we checked this
* earlier.
*/
- if (indexForm->indisclustered)
+ if (thisIndexOid != indexOid)
{
- indexForm->indisclustered = false;
- CatalogTupleUpdate(pg_index, &indexTuple->t_self, indexTuple);
+ Oid parentind = thisIndexOid;
+ set_indisclustered(thisIndexOid, false, pg_index);
+
+ /*
+ * When setting a given index as clustered, also remove
+ * indisclustered from all parents of other partitioned indexes
+ */
+ while (get_rel_relispartition(parentind))
+ {
+ parentind = get_partition_parent(parentind);
+ set_indisclustered(parentind, false, pg_index);
+ }
}
- else if (thisIndexOid == indexOid)
+ else
{
- /* this was checked earlier, but let's be real sure */
- if (!indexForm->indisvalid)
- elog(ERROR, "cannot cluster on invalid index %u", indexOid);
- indexForm->indisclustered = true;
- CatalogTupleUpdate(pg_index, &indexTuple->t_self, indexTuple);
+ set_indisclustered(thisIndexOid, true, pg_index);
}
InvokeObjectPostAlterHookArg(IndexRelationId, thisIndexOid, 0,
InvalidOid, is_internal);
-
- heap_freetuple(indexTuple);
}
table_close(pg_index, RowExclusiveLock);
@@ -565,10 +600,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool isTopLevel, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1563,3 +1594,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, bool isTopLevel, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK,
+ isTopLevel);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9c6f5ecb6a..ad0cd2ec92 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 60c2f45466..d428a94454 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -3202,8 +3202,9 @@ typedef struct AlterSystemStmt
*/
typedef enum ClusterOption
{
- CLUOPT_RECHECK = 1 << 0, /* recheck relation state */
- CLUOPT_VERBOSE = 1 << 1 /* print progress info */
+ CLUOPT_VERBOSE = 1 << 0, /* print progress info */
+ CLUOPT_RECHECK = 1 << 1, /* recheck relation state */
+ CLUOPT_RECHECK_ISCLUSTERED = 1 << 2, /* recheck relation state for indisclustered */
} ClusterOption;
typedef struct ClusterStmt
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..60d2a36255 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,75 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
+-- Check that the parent index is marked not clustered after clustering a partition on a different index:
+CREATE INDEX clstrpart1_idx_2 ON clstrpart1(a);
+CLUSTER clstrpart1 USING clstrpart1_idx_2;
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a)
+Number of partitions: 3 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..93e9c8c6ee 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,34 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+-- Check that the parent index is marked not clustered after clustering a partition on a different index:
+CREATE INDEX clstrpart1_idx_2 ON clstrpart1(a);
+CLUSTER clstrpart1 USING clstrpart1_idx_2;
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FCuugMFkClbJLl1L
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v8-0003-WIP-implement-DROP-INDEX-CONCURRENTLY-on-partitio.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v11] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
VACUUM (including vacuum full) has recursed into partitioned tables since
partitioning were introduced in v10 (3c3bb9933). See expand_vacuum_rel().
For VACUUM FULL, vacuum_rel() calls cluster_rel() for each partition.
This patch is to make cluster() do all the same stuff before calling
cluster_rel().
For now, partitioned indexes cannot be marked clustered, so clustering requires
specification of a partitioned index on the partitioned table.
See also a556549d7 and 19de0ab23.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 177 ++++++++++++++++++++------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 45 ++++++-
src/test/regress/sql/cluster.sql | 21 ++-
6 files changed, 204 insertions(+), 47 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 86f5fdc469b..b3463ae5c46 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -196,6 +196,12 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index (which must be specified).
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 61853e6dec4..196c7f7eeb2 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -131,6 +136,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
Relation rel;
/* Find, lock, and check permissions on the table */
+ /* Obtain AEL now to avoid lock-upgrade hazard in the single-transaction case */
tableOid = RangeVarGetRelidExtended(stmt->relation,
AccessExclusiveLock,
0,
@@ -146,14 +152,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -188,11 +186,50 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
stmt->indexname, stmt->relation->relname)));
}
- /* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* close relation, keep lock till commit */
+ table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /*
+ * Expand partitioned relations for CLUSTER (the corresponding
+ * thing for VACUUM FULL happens in and around expand_vacuum_rel()
+ */
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+
+ /*
+ * For now, partitioned indexes are not actually marked clustered.
+ */
+ check_index_is_clusterable(rel, indexOid, true, AccessShareLock);
+
+ /* close relation, releasing lock on parent table */
+ table_close(rel, AccessExclusiveLock);
+
+ /* Do the job. */
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +239,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +261,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -327,10 +342,11 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/*
* Silently skip a temp table for a remote session. Only doing this
* check in the "recheck" case is appropriate (which currently means
- * somebody is executing a database-wide CLUSTER), because there is
- * another check in cluster() which will stop any attempt to cluster
- * remote temp tables by name. There is another check in cluster_rel
- * which is redundant, but we leave it for extra safety.
+ * somebody is executing a database-wide CLUSTER or on a partitioned
+ * table), because there is another check in cluster() which will stop
+ * any attempt to cluster remote temp tables by name. There is another
+ * check in cluster_rel which is redundant, but we leave it for extra
+ * safety.
*/
if (RELATION_IS_OTHER_TEMP(OldHeap))
{
@@ -352,9 +368,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -415,6 +432,9 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ // Should silently skip this rather than assert ?
+ Assert(RELKIND_HAS_STORAGE(OldHeap->rd_rel->relkind));
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -585,8 +605,8 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
if (OidIsValid(indexOid))
+ /* Mark the correct index as clustered */
mark_index_clustered(OldHeap, indexOid, true);
/* Remember info about rel before closing OldHeap */
@@ -1604,3 +1624,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and their associated indexes, where each index is a
+ * partition of the given index
+ * We're called with a lock held on the parent table.
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ /* Do not lock the children until they're processed. */
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(indexrelid)))
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 39be6f556a8..9944f21f71c 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -599,6 +599,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index 3db375d7cc7..0605b053b61 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index e46a66952f0..3f2758d13f6 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -444,13 +444,52 @@ DROP TABLE clustertest;
CREATE TABLE clustertest (f1 int PRIMARY KEY);
CLUSTER clustertest USING clustertest_pkey;
CLUSTER clustertest;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+CLUSTER clstrpart USING clstrpart_idx;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Partitioned indexes aren't and can't be marked un/clustered:
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a)
+Number of partitions: 3 (Use \d+ to list them.)
+
+CLUSTER clstrpart;
+ERROR: there is no previously clustered index for table "clstrpart"
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+ERROR: cannot mark index clustered in partitioned table
ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
ERROR: cannot mark index clustered in partitioned table
-CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index aee9cf83e04..74118993a82 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -202,11 +202,28 @@ CREATE TABLE clustertest (f1 int PRIMARY KEY);
CLUSTER clustertest USING clustertest_pkey;
CLUSTER clustertest;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Partitioned indexes aren't and can't be marked un/clustered:
+\d clstrpart
+CLUSTER clstrpart;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.1
--NqNl6FRZtoRUn5bW--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v3 1/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 167 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/nodes/parsenodes.h | 5 +-
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 53 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index b9450e7366..0476cfff72 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -172,6 +172,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..391e018bbd 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +74,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -113,7 +118,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -124,14 +129,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +166,32 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +201,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +224,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -328,9 +330,10 @@ cluster_rel(Oid tableOid, Oid indexOid, int options)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -374,8 +377,13 @@ cluster_rel(Oid tableOid, Oid indexOid, int options)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -391,6 +399,14 @@ cluster_rel(Oid tableOid, Oid indexOid, int options)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -459,6 +475,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -483,12 +502,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -560,10 +573,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1557,3 +1566,75 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 5238a960f7..07ef7fc1b7 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 17b4e335c6..c862662a78 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -3198,8 +3198,9 @@ typedef struct AlterSystemStmt
*/
typedef enum ClusterOption
{
- CLUOPT_RECHECK = 1 << 0, /* recheck relation state */
- CLUOPT_VERBOSE = 1 << 1 /* print progress info */
+ CLUOPT_VERBOSE = 1 << 0, /* print progress info */
+ CLUOPT_RECHECK = 1 << 1, /* recheck relation state */
+ CLUOPT_RECHECK_ISCLUSTERED = 1 << 2, /* recheck relation state for indisclustered */
} ClusterOption;
typedef struct ClusterStmt
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--EDJsL2R9iCFAt7IV
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v3-0002-preserve-indisclustered-on-children-of-clustered-.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 167 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/nodes/parsenodes.h | 5 +-
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 53 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index b9450e7366..0476cfff72 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -172,6 +172,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..391e018bbd 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +74,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -113,7 +118,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -124,14 +129,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +166,32 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +201,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +224,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -328,9 +330,10 @@ cluster_rel(Oid tableOid, Oid indexOid, int options)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -374,8 +377,13 @@ cluster_rel(Oid tableOid, Oid indexOid, int options)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -391,6 +399,14 @@ cluster_rel(Oid tableOid, Oid indexOid, int options)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -459,6 +475,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -483,12 +502,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -560,10 +573,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1557,3 +1566,75 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 8afc780acc..1deafd2707 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index d1f9ef29ca..43b6d16a13 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -3198,8 +3198,9 @@ typedef struct AlterSystemStmt
*/
typedef enum ClusterOption
{
- CLUOPT_RECHECK = 1 << 0, /* recheck relation state */
- CLUOPT_VERBOSE = 1 << 1 /* print progress info */
+ CLUOPT_VERBOSE = 1 << 0, /* print progress info */
+ CLUOPT_RECHECK = 1 << 1, /* recheck relation state */
+ CLUOPT_RECHECK_ISCLUSTERED = 1 << 2, /* recheck relation state for indisclustered */
} ClusterOption;
typedef struct ClusterStmt
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--/i8j2F0k9BYX4qLc
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v5-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v3 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index eb018854a5..d6a7ef2c30 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -578,6 +578,7 @@ static const SchemaQuery Query_for_list_of_vacuumables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--SCOJXUq1iwCn05li--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v1 1/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
TODO: handle DB-WIDE "CLUSTER;" for partitioned tables
new partitions need to inherit indisclustered ?
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 169 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/nodes/parsenodes.h | 5 +-
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index b9450e7366..0476cfff72 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -172,6 +172,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 0d647e912c..1db8382a27 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -75,6 +77,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
TransactionId *pFreezeXid,
MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, bool isTopLevel, int options);
/*---------------------------------------------------------------------------
@@ -116,7 +121,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -127,14 +132,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -172,8 +169,32 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options, isTopLevel);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options, isTopLevel);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, isTopLevel, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -183,7 +204,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -207,26 +227,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK,
- isTopLevel);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, isTopLevel, stmt->options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -332,9 +333,10 @@ cluster_rel(Oid tableOid, Oid indexOid, int options, bool isTopLevel)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -378,8 +380,13 @@ cluster_rel(Oid tableOid, Oid indexOid, int options, bool isTopLevel)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -395,6 +402,14 @@ cluster_rel(Oid tableOid, Oid indexOid, int options, bool isTopLevel)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -463,6 +478,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -487,12 +505,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -565,10 +577,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool isTopLevel, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1563,3 +1571,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, bool isTopLevel, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK,
+ isTopLevel);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index b2b4f1fd4d..c3f080e691 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 60c2f45466..d428a94454 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -3202,8 +3202,9 @@ typedef struct AlterSystemStmt
*/
typedef enum ClusterOption
{
- CLUOPT_RECHECK = 1 << 0, /* recheck relation state */
- CLUOPT_VERBOSE = 1 << 1 /* print progress info */
+ CLUOPT_VERBOSE = 1 << 0, /* print progress info */
+ CLUOPT_RECHECK = 1 << 1, /* recheck relation state */
+ CLUOPT_RECHECK_ISCLUSTERED = 1 << 2, /* recheck relation state for indisclustered */
} ClusterOption;
typedef struct ClusterStmt
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--8Tx+BDMK09J610+l
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v1-0002-preserve-indisclustered-on-children-of-clustered-.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
XXX: should mark_index_clustered() SET WITHOUT CLUSTER for any parent indexes
of an index partition which were previously-clustered ?
---
src/backend/commands/cluster.c | 143 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 25 ++++-
src/test/regress/sql/cluster.sql | 14 ++-
4 files changed, 140 insertions(+), 43 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 0d647e912c..0f61c78fb9 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -75,6 +76,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
TransactionId *pFreezeXid,
MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, bool isTopLevel, int options);
/*---------------------------------------------------------------------------
@@ -127,14 +131,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -172,8 +168,39 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options, isTopLevel);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options, isTopLevel);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, isTopLevel, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -183,7 +210,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -207,26 +233,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK,
- isTopLevel);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, isTopLevel, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -487,12 +494,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1563,3 +1564,71 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ MemoryContextSwitchTo(old_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, bool isTopLevel, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK,
+ isTopLevel);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index f41785f11c..49a2648991 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e14e691b05 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,30 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4bed6afbac 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,21 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--MAH+hnPXVZWQ5cD/
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-WIP-implement-DROP-INDEX-CONCURRENTLY-on-partitio.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v10 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a053bc1e45..3fd192b8d2 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index e46a66952f..8fb496434e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -444,14 +444,62 @@ DROP TABLE clustertest;
CREATE TABLE clustertest (f1 int PRIMARY KEY);
CLUSTER clustertest USING clustertest_pkey;
CLUSTER clustertest;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index aee9cf83e0..57fc661863 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -202,12 +202,30 @@ CREATE TABLE clustertest (f1 int PRIMARY KEY);
CLUSTER clustertest USING clustertest_pkey;
CLUSTER clustertest;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--mvpLiMfbWzRoNl4x
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v10-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/2] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
XXX: should mark_index_clustered() SET WITHOUT CLUSTER for any parent indexes
of an index partition which were previously-clustered ?
---
src/backend/commands/cluster.c | 141 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 134 insertions(+), 43 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 0d647e912c..6ead8ffc79 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -75,6 +76,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
TransactionId *pFreezeXid,
MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, bool isTopLevel, int options);
/*---------------------------------------------------------------------------
@@ -127,14 +131,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -172,8 +168,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options, isTopLevel);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options, isTopLevel);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, isTopLevel, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -183,7 +208,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -207,26 +231,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK,
- isTopLevel);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, isTopLevel, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -487,12 +492,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1563,3 +1562,71 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, bool isTopLevel, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK,
+ isTopLevel);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index f41785f11c..49a2648991 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JgQwtEuHJzHdouWu--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v6 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 172 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 208 insertions(+), 54 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..838bcd9e72 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -191,8 +188,32 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 6abcbea963..3a6a4831e1 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..e4448350e7 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+ relname | relkind | ?column?
+-------------+---------+----------
+ clstrpart | p | t
+ clstrpart1 | p | t
+ clstrpart11 | r | f
+ clstrpart12 | p | t
+ clstrpart2 | r | f
+ clstrpart3 | p | t
+ clstrpart33 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+ indexrelid | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..22225dc924 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY 1;
+-- Check that clustering sets new indisclustered:
+SELECT i.indexrelid::regclass::text, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY 1;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--FsscpQKzF/jJk6ya
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v6-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v7 2/7] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index a75647b1cc..781aa95abc 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--AA9g+nFNFPYNJKiL
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v7-0003-Propagate-changes-to-indisclustered-to-child-pare.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v2 2/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 16 ++-
src/test/regress/sql/cluster.sql | 5 +-
4 files changed, 119 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index eb018854a5..d6a7ef2c30 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -578,6 +578,7 @@ static const SchemaQuery Query_for_list_of_vacuumables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..161c09ee42 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,21 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 0
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..db3c271706 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,12 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JwB53PgKC5A7+0Ej
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
filename="v2-0003-Implement-REINDEX-of-partitioned-tables-indexes.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v4 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--JYK4vJDZwFMowpUq--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v8 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
doc/src/sgml/ref/cluster.sgml | 6 +
src/backend/commands/cluster.c | 174 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 5dd21a0189..fb5deddb35 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -192,6 +192,12 @@ CLUSTER [VERBOSE]
are periodically reclustered.
</para>
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ index partition of the given partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..9b6673867c 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,77 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels are also processed by cluster_rel, to
+ * call check_index_is_clusterable() and mark_index_clustered().
+ */
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+
+ MemoryContextSwitchTo(old_context);
+ }
+
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 1e1c315bae..ba301a6b08 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--O5XBE6gyVG5Rl6Rj
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v8-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v5 3/3] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
---
src/backend/commands/cluster.c | 139 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/test/regress/expected/cluster.out | 23 ++++-
src/test/regress/sql/cluster.sql | 12 ++-
4 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 04d12a7ece..7409c17b0b 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -33,6 +33,7 @@
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/progress.h"
@@ -72,6 +73,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -124,14 +128,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -169,8 +165,37 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
/* close relation, keep lock till commit */
table_close(rel, NoLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, stmt->options);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, stmt->options);
+ } else {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Check index directly since cluster_rel isn't called for partitioned table */
+ check_index_is_clusterable(rel, indexOid, true, AccessExclusiveLock);
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, stmt->options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
+ mark_index_clustered(rel, indexOid, true);
+ table_close(rel, NoLock);
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -180,7 +205,6 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -204,25 +228,7 @@ cluster(ClusterStmt *stmt, bool isTopLevel)
*/
rvs = get_tables_to_cluster(cluster_context);
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- stmt->options | CLUOPT_RECHECK);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, stmt->options);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -483,12 +489,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -1557,3 +1557,70 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+
+ MemoryContext old_context = MemoryContextSwitchTo(cluster_context);
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * We have a full list of direct and indirect children, so skip
+ * partitioned tables and just handle their children.
+ */
+ if (get_rel_relkind(relid) == RELKIND_PARTITIONED_TABLE)
+ continue;
+
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ options | CLUOPT_RECHECK);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index c4af40bfa9..e68808218a 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -587,6 +587,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..21b28fde6e 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,13 +439,28 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 2 (Use \d+ to list them.)
+
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..4a76848213 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,11 +196,19 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
CLUSTER clstrpart USING clstrpart_idx;
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
DROP TABLE clstrpart;
-- Test CLUSTER with external tuplesorting
--
2.17.0
--hYooF8G/hrfVAmum--
^ permalink raw reply [nested|flat] 152+ messages in thread
* [PATCH v9 2/8] Implement CLUSTER of partitioned table..
@ 2020-06-07 21:58 Justin Pryzby <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2020-06-07 21:58 UTC (permalink / raw)
This requires either specification of a partitioned index on which to cluster,
or that an partitioned index was previously set clustered.
VACUUM (including vacuum full) has recursed into partition hierarchies since
partitions were introduced in v10 (3c3bb9933).
---
doc/src/sgml/ref/cluster.sgml | 7 ++
src/backend/commands/cluster.c | 173 +++++++++++++++++++-------
src/bin/psql/tab-complete.c | 1 +
src/include/commands/cluster.h | 1 +
src/test/regress/expected/cluster.out | 58 ++++++++-
src/test/regress/sql/cluster.sql | 24 +++-
6 files changed, 209 insertions(+), 55 deletions(-)
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 0d9720fd8e..17509b35eb 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -197,6 +197,13 @@ CLUSTER [VERBOSE]
in the <structname>pg_stat_progress_cluster</structname> view. See
<xref linkend="cluster-progress-reporting"/> for details.
</para>
+
+ <para>
+ Clustering a partitioned table clusters each of its partitions using the
+ partition of the specified partitioned index or (if not specified) the
+ partitioned index marked as clustered.
+ </para>
+
</refsect1>
<refsect1>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 096a06f7b3..0c08ac56dc 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -32,7 +32,9 @@
#include "catalog/index.h"
#include "catalog/namespace.h"
#include "catalog/objectaccess.h"
+#include "catalog/partition.h"
#include "catalog/pg_am.h"
+#include "catalog/pg_inherits.h"
#include "catalog/toasting.h"
#include "commands/cluster.h"
#include "commands/defrem.h"
@@ -73,6 +75,9 @@ static void copy_table_data(Oid OIDNewHeap, Oid OIDOldHeap, Oid OIDOldIndex,
bool verbose, bool *pSwapToastByContent,
TransactionId *pFreezeXid, MultiXactId *pCutoffMulti);
static List *get_tables_to_cluster(MemoryContext cluster_context);
+static List *get_tables_to_cluster_partitioned(MemoryContext cluster_context,
+ Oid indexOid);
+static void cluster_multiple_rels(List *rvs, int options);
/*---------------------------------------------------------------------------
@@ -135,7 +140,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
AccessExclusiveLock,
0,
RangeVarCallbackOwnsTable, NULL);
- rel = table_open(tableOid, NoLock);
+ rel = table_open(tableOid, ShareUpdateExclusiveLock);
/*
* Reject clustering a remote temp table ... their local buffer
@@ -146,14 +151,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
- /*
- * Reject clustering a partitioned table.
- */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot cluster a partitioned table")));
-
if (stmt->indexname == NULL)
{
ListCell *index;
@@ -189,10 +186,34 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
}
/* close relation, keep lock till commit */
- table_close(rel, NoLock);
+ table_close(rel, ShareUpdateExclusiveLock);
- /* Do the job. */
- cluster_rel(tableOid, indexOid, ¶ms);
+ if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+ {
+ /* Do the job. */
+ cluster_rel(tableOid, indexOid, ¶ms);
+ }
+ else
+ {
+ List *rvs;
+ MemoryContext cluster_context;
+
+ /* Refuse to hold strong locks in a user transaction */
+ PreventInTransactionBlock(isTopLevel, "CLUSTER");
+
+ cluster_context = AllocSetContextCreate(PortalContext,
+ "Cluster",
+ ALLOCSET_DEFAULT_SIZES);
+
+ rvs = get_tables_to_cluster_partitioned(cluster_context, indexOid);
+ cluster_multiple_rels(rvs, params.options);
+
+ /* Start a new transaction for the cleanup work. */
+ StartTransactionCommand();
+
+ /* Clean up working storage */
+ MemoryContextDelete(cluster_context);
+ }
}
else
{
@@ -202,7 +223,6 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
*/
MemoryContext cluster_context;
List *rvs;
- ListCell *rv;
/*
* We cannot run this form of CLUSTER inside a user transaction block;
@@ -225,28 +245,7 @@ cluster(ParseState *pstate, ClusterStmt *stmt, bool isTopLevel)
* cluster_context.
*/
rvs = get_tables_to_cluster(cluster_context);
-
- /* Commit to get out of starting transaction */
- PopActiveSnapshot();
- CommitTransactionCommand();
-
- /* Ok, now that we've got them all, cluster them one by one */
- foreach(rv, rvs)
- {
- RelToCluster *rvtc = (RelToCluster *) lfirst(rv);
- ClusterParams cluster_params = params;
-
- /* Start a new transaction for each relation. */
- StartTransactionCommand();
- /* functions in indexes may want a snapshot set */
- PushActiveSnapshot(GetTransactionSnapshot());
- /* Do the job. */
- cluster_params.options |= CLUOPT_RECHECK;
- cluster_rel(rvtc->tableOid, rvtc->indexOid,
- &cluster_params);
- PopActiveSnapshot();
- CommitTransactionCommand();
- }
+ cluster_multiple_rels(rvs, params.options | CLUOPT_RECHECK_ISCLUSTERED);
/* Start a new transaction for the cleanup work. */
StartTransactionCommand();
@@ -352,9 +351,10 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
}
/*
- * Check that the index is still the one with indisclustered set.
+ * Check that the index is still the one with indisclustered set, if needed.
*/
- if (!get_index_isclustered(indexOid))
+ if ((params->options & CLUOPT_RECHECK_ISCLUSTERED) != 0 &&
+ !get_index_isclustered(indexOid))
{
relation_close(OldHeap, AccessExclusiveLock);
pgstat_progress_end_command();
@@ -398,8 +398,13 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
/* Check heap and index are valid to cluster on */
if (OidIsValid(indexOid))
+ {
check_index_is_clusterable(OldHeap, indexOid, recheck, AccessExclusiveLock);
+ /* Mark the index as clustered */
+ mark_index_clustered(OldHeap, indexOid, true);
+ }
+
/*
* Quietly ignore the request if this is a materialized view which has not
* been populated from its query. No harm is done because there is no data
@@ -415,6 +420,14 @@ cluster_rel(Oid tableOid, Oid indexOid, ClusterParams *params)
return;
}
+ /* For a partitioned rel, we're done. */
+ if (!RELKIND_HAS_STORAGE(get_rel_relkind(tableOid)))
+ {
+ relation_close(OldHeap, AccessExclusiveLock);
+ pgstat_progress_end_command();
+ return;
+ }
+
/*
* All predicate locks on the tuples or pages are about to be made
* invalid, because we move tuples around. Promote them to relation
@@ -483,6 +496,9 @@ check_index_is_clusterable(Relation OldHeap, Oid indexOid, bool recheck, LOCKMOD
* the worst consequence of following broken HOT chains would be that we
* might put recently-dead tuples out-of-order in the new table, and there
* is little harm in that.)
+ *
+ * This also refuses to cluster on an "incomplete" partitioned index
+ * created with "ON ONLY".
*/
if (!OldIndex->rd_index->indisvalid)
ereport(ERROR,
@@ -507,12 +523,6 @@ mark_index_clustered(Relation rel, Oid indexOid, bool is_internal)
Relation pg_index;
ListCell *index;
- /* Disallow applying to a partitioned table */
- if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
- ereport(ERROR,
- (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
- errmsg("cannot mark index clustered in partitioned table")));
-
/*
* If the index is already marked clustered, no need to do anything.
*/
@@ -584,10 +594,6 @@ rebuild_relation(Relation OldHeap, Oid indexOid, bool verbose)
TransactionId frozenXid;
MultiXactId cutoffMulti;
- /* Mark the correct index as clustered */
- if (OidIsValid(indexOid))
- mark_index_clustered(OldHeap, indexOid, true);
-
/* Remember info about rel before closing OldHeap */
relpersistence = OldHeap->rd_rel->relpersistence;
is_system_catalog = IsSystemRelation(OldHeap);
@@ -1582,3 +1588,76 @@ get_tables_to_cluster(MemoryContext cluster_context)
return rvs;
}
+
+/*
+ * Return a List of tables and associated index, where each index is a
+ * partition of the given index
+ */
+static List *
+get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
+{
+ List *inhoids;
+ ListCell *lc;
+ List *rvs = NIL;
+ MemoryContext old_context;
+
+ inhoids = find_all_inheritors(indexOid, NoLock, NULL);
+
+ /*
+ * We have to build the list in a different memory context so it will
+ * survive the cross-transaction processing
+ */
+ old_context = MemoryContextSwitchTo(cluster_context);
+
+ foreach(lc, inhoids)
+ {
+ Oid indexrelid = lfirst_oid(lc);
+ Oid relid = IndexGetRelation(indexrelid, false);
+ RelToCluster *rvtc;
+
+ /*
+ * Partitioned rels (including the top indexOid) are included,
+ * so as to be processed by cluster_rel, which calls
+ * check_index_is_clusterable() and mark_index_clustered().
+ */
+ rvtc = (RelToCluster *) palloc(sizeof(RelToCluster));
+ rvtc->tableOid = relid;
+ rvtc->indexOid = indexrelid;
+ rvs = lappend(rvs, rvtc);
+ }
+
+ MemoryContextSwitchTo(old_context);
+ return rvs;
+}
+
+/* Cluster each relation in a separate transaction */
+static void
+cluster_multiple_rels(List *rvs, int options)
+{
+ ListCell *lc;
+
+ /* Commit to get out of starting transaction */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ /* Ok, now that we've got them all, cluster them one by one */
+ foreach(lc, rvs)
+ {
+ RelToCluster *rvtc = (RelToCluster *) lfirst(lc);
+ ClusterParams cluster_params = { .options = options, };
+
+ /* Start a new transaction for each relation. */
+ StartTransactionCommand();
+
+ /* functions in indexes may want a snapshot set */
+ PushActiveSnapshot(GetTransactionSnapshot());
+
+ /* Do the job. */
+ cluster_params.options |= CLUOPT_RECHECK;
+ cluster_rel(rvtc->tableOid, rvtc->indexOid,
+ &cluster_params);
+
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+ }
+}
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 9f0208ac49..53d585c4d3 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -588,6 +588,7 @@ static const SchemaQuery Query_for_list_of_clusterables = {
.catname = "pg_catalog.pg_class c",
.selcondition =
"c.relkind IN (" CppAsString2(RELKIND_RELATION) ", "
+ CppAsString2(RELKIND_PARTITIONED_TABLE) ", "
CppAsString2(RELKIND_MATVIEW) ")",
.viscondition = "pg_catalog.pg_table_is_visible(c.oid)",
.namespace = "c.relnamespace",
diff --git a/src/include/commands/cluster.h b/src/include/commands/cluster.h
index a941f2accd..c30ca01726 100644
--- a/src/include/commands/cluster.h
+++ b/src/include/commands/cluster.h
@@ -22,6 +22,7 @@
/* flag bits for ClusterParams->flags */
#define CLUOPT_RECHECK 0x01 /* recheck relation state */
#define CLUOPT_VERBOSE 0x02 /* print progress info */
+#define CLUOPT_RECHECK_ISCLUSTERED 0x04 /* recheck relation state for indisclustered */
/* options for CLUSTER */
typedef struct ClusterParams
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index bdae8fe00c..c74cfa88cc 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -439,14 +439,62 @@ select * from clstr_temp;
drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+ERROR: cannot cluster on invalid index "clstrpart_only_idx"
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
-ERROR: cannot mark index clustered in partitioned table
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-ERROR: cannot cluster a partitioned table
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+ relname | level | relkind | ?column?
+-------------+-------+---------+----------
+ clstrpart | 0 | p | t
+ clstrpart1 | 1 | p | t
+ clstrpart11 | 2 | r | f
+ clstrpart12 | 2 | p | t
+ clstrpart2 | 1 | r | f
+ clstrpart3 | 1 | p | t
+ clstrpart33 | 2 | r | f
+(7 rows)
+
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+ relname | relkind | indisclustered
+-------------------+---------+----------------
+ clstrpart11_a_idx | i | t
+ clstrpart12_a_idx | I | t
+ clstrpart1_a_idx | I | t
+ clstrpart2_a_idx | i | t
+ clstrpart33_a_idx | i | t
+ clstrpart3_a_idx | I | t
+ clstrpart_idx | I | t
+(7 rows)
+
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
+ Partitioned table "public.clstrpart"
+ Column | Type | Collation | Nullable | Default
+--------+---------+-----------+----------+---------
+ a | integer | | |
+Partition key: RANGE (a)
+Indexes:
+ "clstrpart_idx" btree (a) CLUSTER
+Number of partitions: 3 (Use \d+ to list them.)
+
-- Test CLUSTER with external tuplesorting
create table clstr_4 as select * from tenk1;
create index cluster_sort on clstr_4 (hundred, thousand, tenthous);
diff --git a/src/test/regress/sql/cluster.sql b/src/test/regress/sql/cluster.sql
index 188183647c..9bcc77695c 100644
--- a/src/test/regress/sql/cluster.sql
+++ b/src/test/regress/sql/cluster.sql
@@ -196,12 +196,30 @@ drop table clstr_temp;
RESET SESSION AUTHORIZATION;
--- Check that partitioned tables cannot be clustered
+-- Check that partitioned tables can be clustered
CREATE TABLE clstrpart (a int) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart1 PARTITION OF clstrpart FOR VALUES FROM (1)TO(10) PARTITION BY RANGE (a);
+CREATE TABLE clstrpart11 PARTITION OF clstrpart1 FOR VALUES FROM (1)TO(10);
+CREATE TABLE clstrpart12 PARTITION OF clstrpart1 FOR VALUES FROM (10)TO(20) PARTITION BY RANGE(a);
+CREATE TABLE clstrpart2 PARTITION OF clstrpart FOR VALUES FROM (20)TO(30);
+CREATE TABLE clstrpart3 PARTITION OF clstrpart DEFAULT PARTITION BY RANGE(a);
+CREATE TABLE clstrpart33 PARTITION OF clstrpart3 DEFAULT;
+ALTER TABLE clstrpart SET WITHOUT CLUSTER;
+CREATE INDEX clstrpart_only_idx ON ONLY clstrpart (a);
+CLUSTER clstrpart USING clstrpart_only_idx; -- fails
+DROP INDEX clstrpart_only_idx;
CREATE INDEX clstrpart_idx ON clstrpart (a);
-ALTER TABLE clstrpart CLUSTER ON clstrpart_idx;
+-- Check that clustering sets new relfilenodes:
+CREATE TEMP TABLE old_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
CLUSTER clstrpart USING clstrpart_idx;
-DROP TABLE clstrpart;
+CREATE TEMP TABLE new_cluster_info AS SELECT relname, level, relfilenode, relkind FROM pg_partition_tree('clstrpart'::regclass) AS tree JOIN pg_class c ON c.oid=tree.relid ;
+SELECT relname, old.level, old.relkind, old.relfilenode = new.relfilenode FROM old_cluster_info AS old JOIN new_cluster_info AS new USING (relname) ORDER BY relname COLLATE "C";
+-- Check that clustering sets new indisclustered:
+SELECT relname, relkind, indisclustered FROM pg_partition_tree('clstrpart_idx'::regclass) AS tree JOIN pg_index i ON i.indexrelid=tree.relid JOIN pg_class c ON c.oid=indexrelid ORDER BY relname COLLATE "C";
+CLUSTER clstrpart1 USING clstrpart1_a_idx; -- partition which is itself partitioned
+CLUSTER clstrpart12 USING clstrpart12_a_idx; -- partition which is itself partitioned, no childs
+CLUSTER clstrpart2 USING clstrpart2_a_idx; -- leaf
+\d clstrpart
-- Test CLUSTER with external tuplesorting
--
2.17.0
--fmvA4kSBHQVZhkR6
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment; filename="v9-0003-f-progress-reporting.patch"
^ permalink raw reply [nested|flat] 152+ messages in thread
* Re: Avoiding smgrimmedsync() during nbtree index builds
@ 2022-01-10 22:50 Melanie Plageman <[email protected]>
2022-01-11 17:10 ` Re: Avoiding smgrimmedsync() during nbtree index builds Melanie Plageman <[email protected]>
0 siblings, 1 reply; 152+ messages in thread
From: Melanie Plageman @ 2022-01-10 22:50 UTC (permalink / raw)
To: Andres Freund <[email protected]>; +Cc: Heikki Linnakangas <[email protected]>; pgsql-hackers; Peter Geoghegan <[email protected]>; Thomas Munro <[email protected]>
I have attached a v3 which includes two commits -- one of which
implements the directmgr API and uses it and the other which adds
functionality to use either directmgr or bufmgr API during index build.
Also registering for march commitfest.
On Thu, Dec 9, 2021 at 2:33 PM Andres Freund <[email protected]> wrote:
>
> Hi,
>
> On 2021-11-19 15:11:57 -0500, Melanie Plageman wrote:
> > From 2130175c5d794f60c5f15d6eb1b626c81fb7c39b Mon Sep 17 00:00:00 2001
> > From: Melanie Plageman <[email protected]>
> > Date: Thu, 15 Apr 2021 07:01:01 -0400
> > Subject: [PATCH v2] Index build avoids immed fsync
> >
> > Avoid immediate fsync for just built indexes either by using shared
> > buffers or by leveraging checkpointer's SyncRequest queue. When a
> > checkpoint begins during the index build, if not using shared buffers,
> > the backend will have to do its own fsync.
> > ---
> > src/backend/access/nbtree/nbtree.c | 39 +++---
> > src/backend/access/nbtree/nbtsort.c | 186 +++++++++++++++++++++++-----
> > src/backend/access/transam/xlog.c | 14 +++
> > src/include/access/xlog.h | 1 +
> > 4 files changed, 188 insertions(+), 52 deletions(-)
> >
> > diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c
> > index 40ad0956e0..a2e32f18e6 100644
> > --- a/src/backend/access/nbtree/nbtree.c
> > +++ b/src/backend/access/nbtree/nbtree.c
> > @@ -150,30 +150,29 @@ void
> > btbuildempty(Relation index)
> > {
> > Page metapage;
> > + Buffer metabuf;
> >
> > - /* Construct metapage. */
> > - metapage = (Page) palloc(BLCKSZ);
> > - _bt_initmetapage(metapage, P_NONE, 0, _bt_allequalimage(index, false));
> > -
> > + // TODO: test this.
>
> Shouldn't this path have plenty coverage?
Yep. Sorry.
> > /*
> > - * Write the page and log it. It might seem that an immediate sync would
> > - * be sufficient to guarantee that the file exists on disk, but recovery
> > - * itself might remove it while replaying, for example, an
> > - * XLOG_DBASE_CREATE or XLOG_TBLSPC_CREATE record. Therefore, we need
> > - * this even when wal_level=minimal.
> > + * Construct metapage.
> > + * Because we don't need to lock the relation for extension (since
> > + * noone knows about it yet) and we don't need to initialize the
> > + * new page, as it is done below by _bt_blnewpage(), _bt_getbuf()
> > + * (with P_NEW and BT_WRITE) is overkill.
>
> Isn't the more relevant operation the log_newpage_buffer()?
Returning to this after some time away, many of my comments no longer
make sense to me either. I can't actually tell which diff your question
applies to because this comment was copy-pasted in two different places
in my code. Also, I've removed this comment and added new ones. So,
given all that, is there still something about log_newpage_buffer() I
should be commenting on?
> > However, it might be worth
> > + * either modifying it or adding a new helper function instead of
> > + * calling ReadBufferExtended() directly. We pass mode RBM_ZERO_AND_LOCK
> > + * because we want to hold an exclusive lock on the buffer content
> > */
>
> "modifying it" refers to what?
>
> I don't see a problem using ReadBufferExtended() here. Why would you like to
> replace it with something else?
I would just disregard these comments now.
> > + /*
> > + * Based on the number of tuples, either create a buffered or unbuffered
> > + * write state. if the number of tuples is small, make a buffered write
> > + * if the number of tuples is larger, then we make an unbuffered write state
> > + * and must ensure that we check the redo pointer to know whether or not we
> > + * need to fsync ourselves
> > + */
> >
> > /*
> > * Finish the build by (1) completing the sort of the spool file, (2)
> > * inserting the sorted tuples into btree pages and (3) building the upper
> > * levels. Finally, it may also be necessary to end use of parallelism.
> > */
> > - _bt_leafbuild(buildstate.spool, buildstate.spool2);
> > + if (reltuples > 1000)
>
> I'm ok with some random magic constant here, but it seems worht putting it in
> some constant / #define to make it more obvious.
Done.
> > + _bt_leafbuild(buildstate.spool, buildstate.spool2, false);
> > + else
> > + _bt_leafbuild(buildstate.spool, buildstate.spool2, true);
>
> Why duplicate the function call?
Fixed.
> > /*
> > * allocate workspace for a new, clean btree page, not linked to any siblings.
> > + * If index is not built in shared buffers, buf should be InvalidBuffer
> > */
> > static Page
> > -_bt_blnewpage(uint32 level)
> > +_bt_blnewpage(uint32 level, Buffer buf)
> > {
> > Page page;
> > BTPageOpaque opaque;
> >
> > - page = (Page) palloc(BLCKSZ);
> > + if (buf)
> > + page = BufferGetPage(buf);
> > + else
> > + page = (Page) palloc(BLCKSZ);
> >
> > /* Zero the page and set up standard page header info */
> > _bt_pageinit(page, BLCKSZ);
>
> Ick, that seems pretty ugly API-wise and subsequently too likely to lead to
> pfree()ing a page that's actually in shared buffers. I think it'd make sense
> to separate the allocation from the initialization bits?
Fixed.
> > @@ -635,8 +657,20 @@ _bt_blnewpage(uint32 level)
> > * emit a completed btree page, and release the working storage.
> > */
> > static void
> > -_bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
> > +_bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno, Buffer buf)
> > {
> > + if (wstate->use_shared_buffers)
> > + {
> > + Assert(buf);
> > + START_CRIT_SECTION();
> > + MarkBufferDirty(buf);
> > + if (wstate->btws_use_wal)
> > + log_newpage_buffer(buf, true);
> > + END_CRIT_SECTION();
> > + _bt_relbuf(wstate->index, buf);
> > + return;
> > + }
> > +
> > /* XLOG stuff */
> > if (wstate->btws_use_wal)
> > {
> > @@ -659,7 +693,7 @@ _bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
> > smgrextend(RelationGetSmgr(wstate->index), MAIN_FORKNUM,
> > wstate->btws_pages_written++,
> > (char *) wstate->btws_zeropage,
> > - true);
> > + false);
> > }
>
> Is there a good place to document the way we ensure durability for this path?
I added some new comments. Let me know if you think that I am still
missing this documentation.
> > + /*
> > + * Extend the index relation upfront to reserve the metapage
> > + */
> > + if (wstate->use_shared_buffers)
> > + {
> > + /*
> > + * We should not need to LockRelationForExtension() as no one else knows
> > + * about this index yet?
> > + * Extend the index relation by one block for the metapage. _bt_getbuf()
> > + * is not used here as it does _bt_pageinit() which is one later by
>
> *done
>
>
> > + * _bt_initmetapage(). We will fill in the metapage and write it out at
> > + * the end of index build when we have all of the information required
> > + * for the metapage. However, we initially extend the relation for it to
> > + * occupy block 0 because it is much easier when using shared buffers to
> > + * extend the relation with a block number that is always increasing by
> > + * 1.
>
> Not quite following what you're trying to get at here. There generally is no
> way to extend a relation except by increasing block numbers?
I've updated this comment too. It should make more sense now.
> > @@ -1425,7 +1544,10 @@ _bt_load(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
> > * still not be on disk when the crash occurs.
> > */
> > if (wstate->btws_use_wal)
> > - smgrimmedsync(RelationGetSmgr(wstate->index), MAIN_FORKNUM);
> > + {
> > + if (!wstate->use_shared_buffers && RedoRecPtrChanged(wstate->redo))
> > + smgrimmedsync(RelationGetSmgr(wstate->index), MAIN_FORKNUM);
> > + }
> > }
> >
> > /*
>
> This needs documentation. The whole comment above isn't accurate anymore afaict?
Should be correct now.
> > diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
> > index 1616448368..63fd212787 100644
> > --- a/src/backend/access/transam/xlog.c
> > +++ b/src/backend/access/transam/xlog.c
> > @@ -8704,6 +8704,20 @@ GetRedoRecPtr(void)
> > return RedoRecPtr;
> > }
> >
> > +bool
> > +RedoRecPtrChanged(XLogRecPtr comparator_ptr)
> > +{
> > + XLogRecPtr ptr;
> > +
> > + SpinLockAcquire(&XLogCtl->info_lck);
> > + ptr = XLogCtl->RedoRecPtr;
> > + SpinLockRelease(&XLogCtl->info_lck);
> > +
> > + if (RedoRecPtr < ptr)
> > + RedoRecPtr = ptr;
> > + return RedoRecPtr != comparator_ptr;
> > +}
>
> What's the deal with the < comparison?
I saw that GetRedoRecPtr() does this and thought maybe I should do the
same in this function. I'm not quite sure where I should be getting the
redo pointer.
Maybe I should just call GetRedoRecPtr() and compare it to the one I
saved? I suppose I also thought that maybe someone else in the future
would like to have a function like RedoRecPtrChanged().
- Melanie
Attachments:
[text/x-patch] v3-0001-Add-unbuffered-IO-and-avoid-immed-fsync.patch (27.2K, ../../CAAKRu_ZBLVJ3S04YAOcbmzLtUPvbB+G62QDOUjojvV_My4hUOA@mail.gmail.com/2-v3-0001-Add-unbuffered-IO-and-avoid-immed-fsync.patch)
download | inline diff:
From 492d9d088df670a837efb2c107114457663378d8 Mon Sep 17 00:00:00 2001
From: Melanie Plageman <[email protected]>
Date: Tue, 28 Sep 2021 14:51:11 -0400
Subject: [PATCH v3 1/2] Add unbuffered IO and avoid immed fsync
Replace unbuffered extends and writes
When writing data outside of shared buffers, the backend must do a
series of steps to ensure the data is both durable and recoverable. When
writing or extending a page of data, the backend must log, checksum, and
write out the page before freeing it.
Additionally, though the data is durable once the WAL entries are on
permanent storage, the XLOG Redo pointer cannot be moved past this WAL
until the page data is safely on permanent storage. If a crash were to
occur before the data is fsync'd, the WAL wouldn't be replayed during
recovery, and the data would be lost. Thus, the backend must ensure that
either the Redo pointer has not moved or that the data is fsync'd before
freeing the page.
This is not a problem with pages written in shared buffers because the
checkpointer will block until all buffers that were dirtied before it
began finish before it moves the Redo pointer past their associated WAL
entries.
This commit makes two main changes:
1) It wraps smgrextend() and smgrwrite() in functions from a new API
for writing data outside of shared buffers, directmgr.
2) It saves the XLOG Redo pointer location before doing the write or
extend. It also adds an fsync request for the page to the
checkpointer's pending-ops table. Then, after doing the write or
extend, if the Redo pointer has moved (meaning a checkpoint has
started since it saved it last), then the backend fsync's the page
itself. Otherwise, it lets the checkpointer take care of fsync'ing
the page the next time it processes the pending-ops table.
---
src/backend/access/gist/gistbuild.c | 17 +++----
src/backend/access/hash/hashpage.c | 9 ++--
src/backend/access/heap/heapam_handler.c | 16 ++++---
src/backend/access/heap/rewriteheap.c | 26 ++++-------
src/backend/access/heap/visibilitymap.c | 8 ++--
src/backend/access/nbtree/nbtree.c | 17 ++++---
src/backend/access/nbtree/nbtsort.c | 39 +++++-----------
src/backend/access/spgist/spginsert.c | 25 +++++-----
src/backend/access/transam/xlog.c | 13 ++++++
src/backend/catalog/storage.c | 25 +++-------
src/backend/storage/Makefile | 2 +-
src/backend/storage/direct/Makefile | 17 +++++++
src/backend/storage/direct/directmgr.c | 57 +++++++++++++++++++++++
src/backend/storage/freespace/freespace.c | 10 ++--
src/include/access/xlog.h | 1 +
15 files changed, 175 insertions(+), 107 deletions(-)
create mode 100644 src/backend/storage/direct/Makefile
create mode 100644 src/backend/storage/direct/directmgr.c
diff --git a/src/backend/access/gist/gistbuild.c b/src/backend/access/gist/gistbuild.c
index 9854116fca..748ca65492 100644
--- a/src/backend/access/gist/gistbuild.c
+++ b/src/backend/access/gist/gistbuild.c
@@ -43,6 +43,7 @@
#include "miscadmin.h"
#include "optimizer/optimizer.h"
#include "storage/bufmgr.h"
+#include "storage/directmgr.h"
#include "storage/smgr.h"
#include "utils/memutils.h"
#include "utils/rel.h"
@@ -91,6 +92,7 @@ typedef struct
int64 indtuples; /* number of tuples indexed */
+ UnBufferedWriteState ub_wstate;
/*
* Extra data structures used during a buffering build. 'gfbb' contains
* information related to managing the build buffers. 'parentMap' is a
@@ -194,6 +196,7 @@ gistbuild(Relation heap, Relation index, IndexInfo *indexInfo)
buildstate.heaprel = heap;
buildstate.sortstate = NULL;
buildstate.giststate = initGISTstate(index);
+ buildstate.ub_wstate.smgr_rel = RelationGetSmgr(index);
/*
* Create a temporary memory context that is reset once for each tuple
@@ -403,14 +406,15 @@ gist_indexsortbuild(GISTBuildState *state)
state->pages_allocated = 0;
state->pages_written = 0;
state->ready_num_pages = 0;
+ unbuffered_prep(&state->ub_wstate);
/*
* Write an empty page as a placeholder for the root page. It will be
* replaced with the real root page at the end.
*/
page = palloc0(BLCKSZ);
- smgrextend(RelationGetSmgr(state->indexrel), MAIN_FORKNUM, GIST_ROOT_BLKNO,
- page, true);
+ unbuffered_extend(&state->ub_wstate, MAIN_FORKNUM, GIST_ROOT_BLKNO, page,
+ true);
state->pages_allocated++;
state->pages_written++;
@@ -450,13 +454,12 @@ gist_indexsortbuild(GISTBuildState *state)
/* Write out the root */
PageSetLSN(pagestate->page, GistBuildLSN);
- PageSetChecksumInplace(pagestate->page, GIST_ROOT_BLKNO);
- smgrwrite(RelationGetSmgr(state->indexrel), MAIN_FORKNUM, GIST_ROOT_BLKNO,
- pagestate->page, true);
+ unbuffered_write(&state->ub_wstate, MAIN_FORKNUM, GIST_ROOT_BLKNO, pagestate->page);
if (RelationNeedsWAL(state->indexrel))
log_newpage(&state->indexrel->rd_node, MAIN_FORKNUM, GIST_ROOT_BLKNO,
pagestate->page, true);
+ unbuffered_finish(&state->ub_wstate, MAIN_FORKNUM);
pfree(pagestate->page);
pfree(pagestate);
}
@@ -570,9 +573,7 @@ gist_indexsortbuild_flush_ready_pages(GISTBuildState *state)
elog(ERROR, "unexpected block number to flush GiST sorting build");
PageSetLSN(page, GistBuildLSN);
- PageSetChecksumInplace(page, blkno);
- smgrextend(RelationGetSmgr(state->indexrel), MAIN_FORKNUM, blkno, page,
- true);
+ unbuffered_extend(&state->ub_wstate, MAIN_FORKNUM, blkno, page, false);
state->pages_written++;
}
diff --git a/src/backend/access/hash/hashpage.c b/src/backend/access/hash/hashpage.c
index ee351aea09..722420adf5 100644
--- a/src/backend/access/hash/hashpage.c
+++ b/src/backend/access/hash/hashpage.c
@@ -32,6 +32,7 @@
#include "access/hash_xlog.h"
#include "miscadmin.h"
#include "port/pg_bitutils.h"
+#include "storage/directmgr.h"
#include "storage/lmgr.h"
#include "storage/predicate.h"
#include "storage/smgr.h"
@@ -990,8 +991,10 @@ _hash_alloc_buckets(Relation rel, BlockNumber firstblock, uint32 nblocks)
PGAlignedBlock zerobuf;
Page page;
HashPageOpaque ovflopaque;
+ UnBufferedWriteState ub_wstate;
lastblock = firstblock + nblocks - 1;
+ ub_wstate.smgr_rel = RelationGetSmgr(rel);
/*
* Check for overflow in block number calculation; if so, we cannot extend
@@ -1000,6 +1003,8 @@ _hash_alloc_buckets(Relation rel, BlockNumber firstblock, uint32 nblocks)
if (lastblock < firstblock || lastblock == InvalidBlockNumber)
return false;
+ unbuffered_prep(&ub_wstate);
+
page = (Page) zerobuf.data;
/*
@@ -1024,9 +1029,7 @@ _hash_alloc_buckets(Relation rel, BlockNumber firstblock, uint32 nblocks)
zerobuf.data,
true);
- PageSetChecksumInplace(page, lastblock);
- smgrextend(RelationGetSmgr(rel), MAIN_FORKNUM, lastblock, zerobuf.data,
- false);
+ unbuffered_extend(&ub_wstate, MAIN_FORKNUM, lastblock, zerobuf.data, false);
return true;
}
diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c
index 39ef8a0b77..8824e39a91 100644
--- a/src/backend/access/heap/heapam_handler.c
+++ b/src/backend/access/heap/heapam_handler.c
@@ -38,6 +38,7 @@
#include "pgstat.h"
#include "storage/bufmgr.h"
#include "storage/bufpage.h"
+#include "storage/directmgr.h"
#include "storage/lmgr.h"
#include "storage/predicate.h"
#include "storage/procarray.h"
@@ -575,6 +576,7 @@ heapam_relation_set_new_filenode(Relation rel,
MultiXactId *minmulti)
{
SMgrRelation srel;
+ UnBufferedWriteState ub_wstate;
/*
* Initialize to the minimum XID that could put tuples in the table. We
@@ -594,15 +596,15 @@ heapam_relation_set_new_filenode(Relation rel,
*minmulti = GetOldestMultiXactId();
srel = RelationCreateStorage(*newrnode, persistence);
+ ub_wstate.smgr_rel = srel;
+ unbuffered_prep(&ub_wstate);
/*
* If required, set up an init fork for an unlogged table so that it can
- * be correctly reinitialized on restart. An immediate sync is required
- * even if the page has been logged, because the write did not go through
- * shared_buffers and therefore a concurrent checkpoint may have moved the
- * redo pointer past our xlog record. Recovery may as well remove it
- * while replaying, for example, XLOG_DBASE_CREATE or XLOG_TBLSPC_CREATE
- * record. Therefore, logging is necessary even if wal_level=minimal.
+ * be correctly reinitialized on restart.
+ * Recovery may as well remove our xlog record while replaying, for
+ * example, XLOG_DBASE_CREATE or XLOG_TBLSPC_CREATE record. Therefore,
+ * logging is necessary even if wal_level=minimal.
*/
if (persistence == RELPERSISTENCE_UNLOGGED)
{
@@ -611,7 +613,7 @@ heapam_relation_set_new_filenode(Relation rel,
rel->rd_rel->relkind == RELKIND_TOASTVALUE);
smgrcreate(srel, INIT_FORKNUM, false);
log_smgrcreate(newrnode, INIT_FORKNUM);
- smgrimmedsync(srel, INIT_FORKNUM);
+ unbuffered_finish(&ub_wstate, INIT_FORKNUM);
}
smgrclose(srel);
diff --git a/src/backend/access/heap/rewriteheap.c b/src/backend/access/heap/rewriteheap.c
index 5bb5ebba23..5d93850ccc 100644
--- a/src/backend/access/heap/rewriteheap.c
+++ b/src/backend/access/heap/rewriteheap.c
@@ -119,6 +119,7 @@
#include "replication/logical.h"
#include "replication/slot.h"
#include "storage/bufmgr.h"
+#include "storage/directmgr.h"
#include "storage/fd.h"
#include "storage/procarray.h"
#include "storage/smgr.h"
@@ -152,6 +153,7 @@ typedef struct RewriteStateData
HTAB *rs_old_new_tid_map; /* unmatched B tuples */
HTAB *rs_logical_mappings; /* logical remapping files */
uint32 rs_num_rewrite_mappings; /* # in memory mappings */
+ UnBufferedWriteState rs_unbuffered_wstate;
} RewriteStateData;
/*
@@ -264,6 +266,9 @@ begin_heap_rewrite(Relation old_heap, Relation new_heap, TransactionId oldest_xm
state->rs_freeze_xid = freeze_xid;
state->rs_cutoff_multi = cutoff_multi;
state->rs_cxt = rw_cxt;
+ state->rs_unbuffered_wstate.smgr_rel = RelationGetSmgr(state->rs_new_rel);
+
+ unbuffered_prep(&state->rs_unbuffered_wstate);
/* Initialize hash tables used to track update chains */
hash_ctl.keysize = sizeof(TidHashKey);
@@ -324,21 +329,12 @@ end_heap_rewrite(RewriteState state)
state->rs_buffer,
true);
- PageSetChecksumInplace(state->rs_buffer, state->rs_blockno);
-
- smgrextend(RelationGetSmgr(state->rs_new_rel), MAIN_FORKNUM,
- state->rs_blockno, (char *) state->rs_buffer, true);
+ unbuffered_extend(&state->rs_unbuffered_wstate, MAIN_FORKNUM,
+ state->rs_blockno, state->rs_buffer, false);
}
- /*
- * When we WAL-logged rel pages, we must nonetheless fsync them. The
- * reason is the same as in storage.c's RelationCopyStorage(): we're
- * writing data that's not in shared buffers, and so a CHECKPOINT
- * occurring during the rewriteheap operation won't have fsync'd data we
- * wrote before the checkpoint.
- */
if (RelationNeedsWAL(state->rs_new_rel))
- smgrimmedsync(RelationGetSmgr(state->rs_new_rel), MAIN_FORKNUM);
+ unbuffered_finish(&state->rs_unbuffered_wstate, MAIN_FORKNUM);
logical_end_heap_rewrite(state);
@@ -690,10 +686,8 @@ raw_heap_insert(RewriteState state, HeapTuple tup)
* need for smgr to schedule an fsync for this write; we'll do it
* ourselves in end_heap_rewrite.
*/
- PageSetChecksumInplace(page, state->rs_blockno);
-
- smgrextend(RelationGetSmgr(state->rs_new_rel), MAIN_FORKNUM,
- state->rs_blockno, (char *) page, true);
+ unbuffered_extend(&state->rs_unbuffered_wstate, MAIN_FORKNUM,
+ state->rs_blockno, page, false);
state->rs_blockno++;
state->rs_buffer_valid = false;
diff --git a/src/backend/access/heap/visibilitymap.c b/src/backend/access/heap/visibilitymap.c
index 9032d4758f..1a7482e267 100644
--- a/src/backend/access/heap/visibilitymap.c
+++ b/src/backend/access/heap/visibilitymap.c
@@ -92,6 +92,7 @@
#include "miscadmin.h"
#include "port/pg_bitutils.h"
#include "storage/bufmgr.h"
+#include "storage/directmgr.h"
#include "storage/lmgr.h"
#include "storage/smgr.h"
#include "utils/inval.h"
@@ -616,7 +617,9 @@ vm_extend(Relation rel, BlockNumber vm_nblocks)
BlockNumber vm_nblocks_now;
PGAlignedBlock pg;
SMgrRelation reln;
+ UnBufferedWriteState ub_wstate;
+ ub_wstate.smgr_rel = RelationGetSmgr(rel);
PageInit((Page) pg.data, BLCKSZ, 0);
/*
@@ -654,9 +657,8 @@ vm_extend(Relation rel, BlockNumber vm_nblocks)
/* Now extend the file */
while (vm_nblocks_now < vm_nblocks)
{
- PageSetChecksumInplace((Page) pg.data, vm_nblocks_now);
-
- smgrextend(reln, VISIBILITYMAP_FORKNUM, vm_nblocks_now, pg.data, false);
+ // TODO: aren't these pages empty? why checksum them
+ unbuffered_extend(&ub_wstate, VISIBILITYMAP_FORKNUM, vm_nblocks_now, (Page) pg.data, false);
vm_nblocks_now++;
}
diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c
index 13024af2fa..6ab6651420 100644
--- a/src/backend/access/nbtree/nbtree.c
+++ b/src/backend/access/nbtree/nbtree.c
@@ -29,6 +29,7 @@
#include "pgstat.h"
#include "postmaster/autovacuum.h"
#include "storage/condition_variable.h"
+#include "storage/directmgr.h"
#include "storage/indexfsm.h"
#include "storage/ipc.h"
#include "storage/lmgr.h"
@@ -151,6 +152,11 @@ void
btbuildempty(Relation index)
{
Page metapage;
+ UnBufferedWriteState wstate;
+
+ wstate.smgr_rel = RelationGetSmgr(index);
+
+ unbuffered_prep(&wstate);
/* Construct metapage. */
metapage = (Page) palloc(BLCKSZ);
@@ -163,18 +169,15 @@ btbuildempty(Relation index)
* XLOG_DBASE_CREATE or XLOG_TBLSPC_CREATE record. Therefore, we need
* this even when wal_level=minimal.
*/
- PageSetChecksumInplace(metapage, BTREE_METAPAGE);
- smgrwrite(RelationGetSmgr(index), INIT_FORKNUM, BTREE_METAPAGE,
- (char *) metapage, true);
+ unbuffered_write(&wstate, INIT_FORKNUM, BTREE_METAPAGE, metapage);
log_newpage(&RelationGetSmgr(index)->smgr_rnode.node, INIT_FORKNUM,
BTREE_METAPAGE, metapage, true);
/*
- * An immediate sync is required even if we xlog'd the page, because the
- * write did not go through shared_buffers and therefore a concurrent
- * checkpoint may have moved the redo pointer past our xlog record.
+ * Even though we xlog'd the page, a concurrent checkpoint may have moved
+ * the redo pointer past our xlog record, so we may still need to fsync.
*/
- smgrimmedsync(RelationGetSmgr(index), INIT_FORKNUM);
+ unbuffered_finish(&wstate, INIT_FORKNUM);
}
/*
diff --git a/src/backend/access/nbtree/nbtsort.c b/src/backend/access/nbtree/nbtsort.c
index dc220146fd..5687acd99d 100644
--- a/src/backend/access/nbtree/nbtsort.c
+++ b/src/backend/access/nbtree/nbtsort.c
@@ -57,6 +57,7 @@
#include "executor/instrument.h"
#include "miscadmin.h"
#include "pgstat.h"
+#include "storage/directmgr.h"
#include "storage/smgr.h"
#include "tcop/tcopprot.h" /* pgrminclude ignore */
#include "utils/rel.h"
@@ -253,6 +254,7 @@ typedef struct BTWriteState
BlockNumber btws_pages_alloced; /* # pages allocated */
BlockNumber btws_pages_written; /* # pages written out */
Page btws_zeropage; /* workspace for filling zeroes */
+ UnBufferedWriteState ub_wstate;
} BTWriteState;
@@ -560,6 +562,8 @@ _bt_leafbuild(BTSpool *btspool, BTSpool *btspool2)
wstate.heap = btspool->heap;
wstate.index = btspool->index;
+ wstate.ub_wstate.smgr_rel = RelationGetSmgr(btspool->index);
+ wstate.ub_wstate.redo = InvalidXLogRecPtr;
wstate.inskey = _bt_mkscankey(wstate.index, NULL);
/* _bt_mkscankey() won't set allequalimage without metapage */
wstate.inskey->allequalimage = _bt_allequalimage(wstate.index, true);
@@ -656,31 +660,19 @@ _bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
if (!wstate->btws_zeropage)
wstate->btws_zeropage = (Page) palloc0(BLCKSZ);
/* don't set checksum for all-zero page */
- smgrextend(RelationGetSmgr(wstate->index), MAIN_FORKNUM,
- wstate->btws_pages_written++,
- (char *) wstate->btws_zeropage,
- true);
+ unbuffered_extend(&wstate->ub_wstate, MAIN_FORKNUM, wstate->btws_pages_written++, wstate->btws_zeropage, true);
}
- PageSetChecksumInplace(page, blkno);
- /*
- * Now write the page. There's no need for smgr to schedule an fsync for
- * this write; we'll do it ourselves before ending the build.
- */
+ /* Now write the page. Either we are extending the file... */
if (blkno == wstate->btws_pages_written)
{
- /* extending the file... */
- smgrextend(RelationGetSmgr(wstate->index), MAIN_FORKNUM, blkno,
- (char *) page, true);
+ unbuffered_extend(&wstate->ub_wstate, MAIN_FORKNUM, blkno, page, false);
wstate->btws_pages_written++;
}
+ /* or we are overwriting a block we zero-filled before. */
else
- {
- /* overwriting a block we zero-filled before */
- smgrwrite(RelationGetSmgr(wstate->index), MAIN_FORKNUM, blkno,
- (char *) page, true);
- }
+ unbuffered_write(&wstate->ub_wstate, MAIN_FORKNUM, blkno, page);
pfree(page);
}
@@ -1189,6 +1181,8 @@ _bt_load(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
int64 tuples_done = 0;
bool deduplicate;
+
+ unbuffered_prep(&wstate->ub_wstate);
deduplicate = wstate->inskey->allequalimage && !btspool->isunique &&
BTGetDeduplicateItems(wstate->index);
@@ -1415,17 +1409,8 @@ _bt_load(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
/* Close down final pages and write the metapage */
_bt_uppershutdown(wstate, state);
- /*
- * When we WAL-logged index pages, we must nonetheless fsync index files.
- * Since we're building outside shared buffers, a CHECKPOINT occurring
- * during the build has no way to flush the previously written data to
- * disk (indeed it won't know the index even exists). A crash later on
- * would replay WAL from the checkpoint, therefore it wouldn't replay our
- * earlier WAL entries. If we do not fsync those pages here, they might
- * still not be on disk when the crash occurs.
- */
if (wstate->btws_use_wal)
- smgrimmedsync(RelationGetSmgr(wstate->index), MAIN_FORKNUM);
+ unbuffered_finish(&wstate->ub_wstate, MAIN_FORKNUM);
}
/*
diff --git a/src/backend/access/spgist/spginsert.c b/src/backend/access/spgist/spginsert.c
index bfb74049d0..9fd5686b8d 100644
--- a/src/backend/access/spgist/spginsert.c
+++ b/src/backend/access/spgist/spginsert.c
@@ -25,6 +25,7 @@
#include "catalog/index.h"
#include "miscadmin.h"
#include "storage/bufmgr.h"
+#include "storage/directmgr.h"
#include "storage/smgr.h"
#include "utils/memutils.h"
#include "utils/rel.h"
@@ -156,6 +157,10 @@ void
spgbuildempty(Relation index)
{
Page page;
+ UnBufferedWriteState wstate;
+
+ wstate.smgr_rel = RelationGetSmgr(index);
+ unbuffered_prep(&wstate);
/* Construct metapage. */
page = (Page) palloc(BLCKSZ);
@@ -168,36 +173,30 @@ spgbuildempty(Relation index)
* of their existing content when the corresponding create records are
* replayed.
*/
- PageSetChecksumInplace(page, SPGIST_METAPAGE_BLKNO);
- smgrwrite(RelationGetSmgr(index), INIT_FORKNUM, SPGIST_METAPAGE_BLKNO,
- (char *) page, true);
+ unbuffered_write(&wstate, INIT_FORKNUM, SPGIST_METAPAGE_BLKNO, page);
log_newpage(&(RelationGetSmgr(index))->smgr_rnode.node, INIT_FORKNUM,
SPGIST_METAPAGE_BLKNO, page, true);
/* Likewise for the root page. */
SpGistInitPage(page, SPGIST_LEAF);
- PageSetChecksumInplace(page, SPGIST_ROOT_BLKNO);
- smgrwrite(RelationGetSmgr(index), INIT_FORKNUM, SPGIST_ROOT_BLKNO,
- (char *) page, true);
+ unbuffered_write(&wstate, INIT_FORKNUM, SPGIST_ROOT_BLKNO, page);
log_newpage(&(RelationGetSmgr(index))->smgr_rnode.node, INIT_FORKNUM,
SPGIST_ROOT_BLKNO, page, true);
/* Likewise for the null-tuples root page. */
SpGistInitPage(page, SPGIST_LEAF | SPGIST_NULLS);
- PageSetChecksumInplace(page, SPGIST_NULL_BLKNO);
- smgrwrite(RelationGetSmgr(index), INIT_FORKNUM, SPGIST_NULL_BLKNO,
- (char *) page, true);
+ unbuffered_write(&wstate, INIT_FORKNUM, SPGIST_NULL_BLKNO, page);
log_newpage(&(RelationGetSmgr(index))->smgr_rnode.node, INIT_FORKNUM,
SPGIST_NULL_BLKNO, page, true);
/*
- * An immediate sync is required even if we xlog'd the pages, because the
- * writes did not go through shared buffers and therefore a concurrent
- * checkpoint may have moved the redo pointer past our xlog record.
+ * Because the writes did not go through shared buffers, if a concurrent
+ * checkpoint moved the redo pointer past our xlog record, an immediate
+ * sync is required even if we xlog'd the pages.
*/
- smgrimmedsync(RelationGetSmgr(index), INIT_FORKNUM);
+ unbuffered_finish(&wstate, INIT_FORKNUM);
}
/*
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index c9d4cbf3ff..a06e8d005b 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -13191,6 +13191,19 @@ CheckForStandbyTrigger(void)
return false;
}
+bool RedoRecPtrChanged(XLogRecPtr comparator_ptr)
+{
+ XLogRecPtr ptr;
+ SpinLockAcquire(&XLogCtl->info_lck);
+ ptr = XLogCtl->RedoRecPtr;
+ SpinLockRelease(&XLogCtl->info_lck);
+
+ if (RedoRecPtr < ptr)
+ RedoRecPtr = ptr;
+
+ return RedoRecPtr != comparator_ptr;
+}
+
/*
* Remove the files signaling a standby promotion request.
*/
diff --git a/src/backend/catalog/storage.c b/src/backend/catalog/storage.c
index 9b8075536a..05e66d0434 100644
--- a/src/backend/catalog/storage.c
+++ b/src/backend/catalog/storage.c
@@ -28,6 +28,7 @@
#include "catalog/storage.h"
#include "catalog/storage_xlog.h"
#include "miscadmin.h"
+#include "storage/directmgr.h"
#include "storage/freespace.h"
#include "storage/smgr.h"
#include "utils/hsearch.h"
@@ -420,6 +421,10 @@ RelationCopyStorage(SMgrRelation src, SMgrRelation dst,
bool copying_initfork;
BlockNumber nblocks;
BlockNumber blkno;
+ UnBufferedWriteState wstate;
+
+ wstate.smgr_rel = dst;
+ unbuffered_prep(&wstate);
page = (Page) buf.data;
@@ -477,27 +482,11 @@ RelationCopyStorage(SMgrRelation src, SMgrRelation dst,
if (use_wal)
log_newpage(&dst->smgr_rnode.node, forkNum, blkno, page, false);
- PageSetChecksumInplace(page, blkno);
-
- /*
- * Now write the page. We say skipFsync = true because there's no
- * need for smgr to schedule an fsync for this write; we'll do it
- * ourselves below.
- */
- smgrextend(dst, forkNum, blkno, buf.data, true);
+ unbuffered_extend(&wstate, forkNum, blkno, buf.data, false);
}
- /*
- * When we WAL-logged rel pages, we must nonetheless fsync them. The
- * reason is that since we're copying outside shared buffers, a CHECKPOINT
- * occurring during the copy has no way to flush the previously written
- * data to disk (indeed it won't know the new rel even exists). A crash
- * later on would replay WAL from the checkpoint, therefore it wouldn't
- * replay our earlier WAL entries. If we do not fsync those pages here,
- * they might still not be on disk when the crash occurs.
- */
if (use_wal || copying_initfork)
- smgrimmedsync(dst, forkNum);
+ unbuffered_finish(&wstate, forkNum);
}
/*
diff --git a/src/backend/storage/Makefile b/src/backend/storage/Makefile
index 8376cdfca2..501fae5f9d 100644
--- a/src/backend/storage/Makefile
+++ b/src/backend/storage/Makefile
@@ -8,6 +8,6 @@ subdir = src/backend/storage
top_builddir = ../../..
include $(top_builddir)/src/Makefile.global
-SUBDIRS = buffer file freespace ipc large_object lmgr page smgr sync
+SUBDIRS = buffer direct file freespace ipc large_object lmgr page smgr sync
include $(top_srcdir)/src/backend/common.mk
diff --git a/src/backend/storage/direct/Makefile b/src/backend/storage/direct/Makefile
new file mode 100644
index 0000000000..d82bbed48c
--- /dev/null
+++ b/src/backend/storage/direct/Makefile
@@ -0,0 +1,17 @@
+#-------------------------------------------------------------------------
+#
+# Makefile--
+# Makefile for storage/direct
+#
+# IDENTIFICATION
+# src/backend/storage/direct/Makefile
+#
+#-------------------------------------------------------------------------
+
+subdir = src/backend/storage/direct
+top_builddir = ../../../..
+include $(top_builddir)/src/Makefile.global
+
+OBJS = directmgr.o
+
+include $(top_srcdir)/src/backend/common.mk
diff --git a/src/backend/storage/direct/directmgr.c b/src/backend/storage/direct/directmgr.c
new file mode 100644
index 0000000000..d33c2a7a5f
--- /dev/null
+++ b/src/backend/storage/direct/directmgr.c
@@ -0,0 +1,57 @@
+/*-------------------------------------------------------------------------
+ *
+ * directmgr.c
+ * routines for managing unbuffered IO
+ *
+ * Portions Copyright (c) 1996-2021, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ *
+ * IDENTIFICATION
+ * src/backend/storage/direct/directmgr.c
+ *
+ *-------------------------------------------------------------------------
+ */
+#include "postgres.h"
+
+
+#include "access/xlog.h"
+#include "storage/directmgr.h"
+#include "utils/rel.h"
+
+void unbuffered_prep(UnBufferedWriteState *wstate)
+{
+ wstate->redo = GetRedoRecPtr();
+}
+
+/*
+ * When writing data outside shared buffers, a concurrent CHECKPOINT can move
+ * the redo pointer past our WAL entries and won't flush our data to disk. If
+ * the database crashes before the data makes it to disk, our WAL won't be
+ * replayed and the data will be lost.
+ * Thus, if a CHECKPOINT begins between unbuffered_prep() and
+ * unbuffered_finish(), the backend must fsync the data itself.
+ */
+void unbuffered_finish(UnBufferedWriteState *wstate, ForkNumber forknum)
+{
+ if (RedoRecPtrChanged(wstate->redo))
+ smgrimmedsync(wstate->smgr_rel, forknum);
+}
+
+void
+unbuffered_write(UnBufferedWriteState *wstate, ForkNumber forknum, BlockNumber
+ blocknum, Page page)
+{
+ PageSetChecksumInplace(page, blocknum);
+ smgrwrite(wstate->smgr_rel, forknum, blocknum, (char *) page, false);
+}
+
+void
+unbuffered_extend(UnBufferedWriteState *wstate, ForkNumber forknum, BlockNumber
+ blocknum, Page page, bool empty)
+{
+ if (!empty)
+ PageSetChecksumInplace(page, blocknum);
+ smgrextend(wstate->smgr_rel, forknum, blocknum, (char *) page, false);
+}
+
diff --git a/src/backend/storage/freespace/freespace.c b/src/backend/storage/freespace/freespace.c
index c88cb91f06..d31f75bd60 100644
--- a/src/backend/storage/freespace/freespace.c
+++ b/src/backend/storage/freespace/freespace.c
@@ -26,6 +26,7 @@
#include "access/htup_details.h"
#include "access/xlogutils.h"
#include "miscadmin.h"
+#include "storage/directmgr.h"
#include "storage/freespace.h"
#include "storage/fsm_internals.h"
#include "storage/lmgr.h"
@@ -608,6 +609,9 @@ fsm_extend(Relation rel, BlockNumber fsm_nblocks)
BlockNumber fsm_nblocks_now;
PGAlignedBlock pg;
SMgrRelation reln;
+ UnBufferedWriteState ub_wstate;
+
+ ub_wstate.smgr_rel = RelationGetSmgr(rel);
PageInit((Page) pg.data, BLCKSZ, 0);
@@ -647,10 +651,8 @@ fsm_extend(Relation rel, BlockNumber fsm_nblocks)
/* Extend as needed. */
while (fsm_nblocks_now < fsm_nblocks)
{
- PageSetChecksumInplace((Page) pg.data, fsm_nblocks_now);
-
- smgrextend(reln, FSM_FORKNUM, fsm_nblocks_now,
- pg.data, false);
+ // TODO: why was it checksumming all zero pages?
+ unbuffered_extend(&ub_wstate, FSM_FORKNUM, fsm_nblocks_now, (Page) pg.data, false);
fsm_nblocks_now++;
}
diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h
index bb0c52686a..4961067f81 100644
--- a/src/include/access/xlog.h
+++ b/src/include/access/xlog.h
@@ -312,6 +312,7 @@ extern XLogRecPtr GetInsertRecPtr(void);
extern XLogRecPtr GetFlushRecPtr(TimeLineID *insertTLI);
extern TimeLineID GetWALInsertionTimeLine(void);
extern XLogRecPtr GetLastImportantRecPtr(void);
+extern bool RedoRecPtrChanged(XLogRecPtr comparator_ptr);
extern void RemovePromoteSignalFiles(void);
extern bool PromoteIsTriggered(void);
--
2.30.2
[text/x-patch] v3-0002-Use-shared-buffers-when-possible-for-index-build.patch (19.4K, ../../CAAKRu_ZBLVJ3S04YAOcbmzLtUPvbB+G62QDOUjojvV_My4hUOA@mail.gmail.com/3-v3-0002-Use-shared-buffers-when-possible-for-index-build.patch)
download | inline diff:
From f1bd40b2880d9cffeb2e313719c3fd6b4c8e5f04 Mon Sep 17 00:00:00 2001
From: Melanie Plageman <[email protected]>
Date: Mon, 10 Jan 2022 17:34:01 -0500
Subject: [PATCH v3 2/2] Use shared buffers when possible for index build
When there are not too many tuples, building the index in shared buffers
makes sense. It allows the buffer manager to handle how best to do the
IO.
---
src/backend/access/nbtree/nbtree.c | 34 ++--
src/backend/access/nbtree/nbtsort.c | 268 ++++++++++++++++++++++------
2 files changed, 225 insertions(+), 77 deletions(-)
diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c
index 6ab6651420..0714c7fdd6 100644
--- a/src/backend/access/nbtree/nbtree.c
+++ b/src/backend/access/nbtree/nbtree.c
@@ -151,33 +151,27 @@ bthandler(PG_FUNCTION_ARGS)
void
btbuildempty(Relation index)
{
+ /*
+ * Since this only writes one page, use shared buffers.
+ */
Page metapage;
- UnBufferedWriteState wstate;
-
- wstate.smgr_rel = RelationGetSmgr(index);
-
- unbuffered_prep(&wstate);
-
- /* Construct metapage. */
- metapage = (Page) palloc(BLCKSZ);
- _bt_initmetapage(metapage, P_NONE, 0, _bt_allequalimage(index, false));
+ Buffer metabuf;
/*
- * Write the page and log it. It might seem that an immediate sync would
- * be sufficient to guarantee that the file exists on disk, but recovery
- * itself might remove it while replaying, for example, an
- * XLOG_DBASE_CREATE or XLOG_TBLSPC_CREATE record. Therefore, we need
- * this even when wal_level=minimal.
+ * Allocate a buffer for metapage and initialize metapage.
*/
- unbuffered_write(&wstate, INIT_FORKNUM, BTREE_METAPAGE, metapage);
- log_newpage(&RelationGetSmgr(index)->smgr_rnode.node, INIT_FORKNUM,
- BTREE_METAPAGE, metapage, true);
+ metabuf = ReadBufferExtended(index, INIT_FORKNUM, P_NEW, RBM_ZERO_AND_LOCK, NULL);
+ metapage = BufferGetPage(metabuf);
+ _bt_initmetapage(metapage, P_NONE, 0, _bt_allequalimage(index, false));
/*
- * Even though we xlog'd the page, a concurrent checkpoint may have moved
- * the redo pointer past our xlog record, so we may still need to fsync.
+ * Mark metapage buffer as dirty and XLOG it
*/
- unbuffered_finish(&wstate, INIT_FORKNUM);
+ START_CRIT_SECTION();
+ MarkBufferDirty(metabuf);
+ log_newpage_buffer(metabuf, true);
+ END_CRIT_SECTION();
+ _bt_relbuf(index, metabuf);
}
/*
diff --git a/src/backend/access/nbtree/nbtsort.c b/src/backend/access/nbtree/nbtsort.c
index 5687acd99d..68b53e6c04 100644
--- a/src/backend/access/nbtree/nbtsort.c
+++ b/src/backend/access/nbtree/nbtsort.c
@@ -234,6 +234,7 @@ typedef struct BTPageState
{
Page btps_page; /* workspace for page building */
BlockNumber btps_blkno; /* block # to write this page at */
+ Buffer btps_buf; /* buffer to write this page to */
IndexTuple btps_lowkey; /* page's strict lower bound pivot tuple */
OffsetNumber btps_lastoff; /* last item offset loaded */
Size btps_lastextra; /* last item's extra posting list space */
@@ -251,10 +252,25 @@ typedef struct BTWriteState
Relation index;
BTScanInsert inskey; /* generic insertion scankey */
bool btws_use_wal; /* dump pages to WAL? */
- BlockNumber btws_pages_alloced; /* # pages allocated */
- BlockNumber btws_pages_written; /* # pages written out */
+ BlockNumber btws_pages_alloced; /* # pages allocated for index builds outside SB */
+ BlockNumber btws_pages_written; /* # pages written out for index builds outside SB */
Page btws_zeropage; /* workspace for filling zeroes */
UnBufferedWriteState ub_wstate;
+ /*
+ * Allocate a new btree page. This does not initialize the page.
+ */
+ Page (*_bt_bl_alloc_page) (struct BTWriteState *wstate, BlockNumber
+ *blockno, Buffer *buf);
+ /*
+ * Emit a completed btree page, and release the working storage.
+ */
+ void (*_bt_blwritepage) (struct BTWriteState *wstate, Page page,
+ BlockNumber blkno, Buffer buf);
+
+ void (*_bt_bl_unbuffered_prep) (UnBufferedWriteState *wstate);
+
+ void (*_bt_bl_unbuffered_finish) (UnBufferedWriteState *wstate, ForkNumber
+ forknum);
} BTWriteState;
@@ -263,10 +279,22 @@ static double _bt_spools_heapscan(Relation heap, Relation index,
static void _bt_spooldestroy(BTSpool *btspool);
static void _bt_spool(BTSpool *btspool, ItemPointer self,
Datum *values, bool *isnull);
-static void _bt_leafbuild(BTSpool *btspool, BTSpool *btspool2);
+static void _bt_leafbuild(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2);
static void _bt_build_callback(Relation index, ItemPointer tid, Datum *values,
bool *isnull, bool tupleIsAlive, void *state);
-static Page _bt_blnewpage(uint32 level);
+
+static Page
+_bt_bl_alloc_page_direct(BTWriteState *wstate, BlockNumber *blockno, Buffer *buf);
+static Page
+_bt_bl_alloc_page_shared(BTWriteState *wstate, BlockNumber *blockno, Buffer *buf);
+
+static Page _bt_blnewpage(uint32 level, Page page);
+
+static void
+_bt_blwritepage_direct(BTWriteState *wstate, Page page, BlockNumber blkno, Buffer buf);
+static void
+_bt_blwritepage_shared(BTWriteState *wstate, Page page, BlockNumber blkno, Buffer buf);
+
static BTPageState *_bt_pagestate(BTWriteState *wstate, uint32 level);
static void _bt_slideleft(Page rightmostpage);
static void _bt_sortaddtup(Page page, Size itemsize,
@@ -277,9 +305,10 @@ static void _bt_buildadd(BTWriteState *wstate, BTPageState *state,
static void _bt_sort_dedup_finish_pending(BTWriteState *wstate,
BTPageState *state,
BTDedupState dstate);
-static void _bt_uppershutdown(BTWriteState *wstate, BTPageState *state);
static void _bt_load(BTWriteState *wstate,
BTSpool *btspool, BTSpool *btspool2);
+static void _bt_uppershutdown(BTWriteState *wstate, BTPageState *state, Buffer
+ metabuf, Page metapage);
static void _bt_begin_parallel(BTBuildState *buildstate, bool isconcurrent,
int request);
static void _bt_end_parallel(BTLeader *btleader);
@@ -292,6 +321,21 @@ static void _bt_parallel_scan_and_sort(BTSpool *btspool, BTSpool *btspool2,
Sharedsort *sharedsort2, int sortmem,
bool progress);
+#define BT_BUILD_SB_THRESHOLD 1024
+
+static const BTWriteState wstate_shared = {
+ ._bt_bl_alloc_page = _bt_bl_alloc_page_shared,
+ ._bt_blwritepage = _bt_blwritepage_shared,
+ ._bt_bl_unbuffered_prep = NULL,
+ ._bt_bl_unbuffered_finish = NULL,
+};
+
+static const BTWriteState wstate_direct = {
+ ._bt_bl_alloc_page = _bt_bl_alloc_page_direct,
+ ._bt_blwritepage = _bt_blwritepage_direct,
+ ._bt_bl_unbuffered_prep = unbuffered_prep,
+ ._bt_bl_unbuffered_finish = unbuffered_finish,
+};
/*
* btbuild() -- build a new btree index.
@@ -301,6 +345,7 @@ btbuild(Relation heap, Relation index, IndexInfo *indexInfo)
{
IndexBuildResult *result;
BTBuildState buildstate;
+ BTWriteState writestate;
double reltuples;
#ifdef BTREE_BUILD_STATS
@@ -330,8 +375,12 @@ btbuild(Relation heap, Relation index, IndexInfo *indexInfo)
* Finish the build by (1) completing the sort of the spool file, (2)
* inserting the sorted tuples into btree pages and (3) building the upper
* levels. Finally, it may also be necessary to end use of parallelism.
+ *
+ * Don't use shared buffers if the number of tuples is too large.
*/
- _bt_leafbuild(buildstate.spool, buildstate.spool2);
+ writestate = reltuples < BT_BUILD_SB_THRESHOLD ? wstate_shared : wstate_direct;
+
+ _bt_leafbuild(&writestate, buildstate.spool, buildstate.spool2);
_bt_spooldestroy(buildstate.spool);
if (buildstate.spool2)
_bt_spooldestroy(buildstate.spool2);
@@ -537,10 +586,8 @@ _bt_spool(BTSpool *btspool, ItemPointer self, Datum *values, bool *isnull)
* create an entire btree.
*/
static void
-_bt_leafbuild(BTSpool *btspool, BTSpool *btspool2)
+_bt_leafbuild(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
{
- BTWriteState wstate;
-
#ifdef BTREE_BUILD_STATS
if (log_btree_build_stats)
{
@@ -560,23 +607,38 @@ _bt_leafbuild(BTSpool *btspool, BTSpool *btspool2)
tuplesort_performsort(btspool2->sortstate);
}
- wstate.heap = btspool->heap;
- wstate.index = btspool->index;
- wstate.ub_wstate.smgr_rel = RelationGetSmgr(btspool->index);
- wstate.ub_wstate.redo = InvalidXLogRecPtr;
- wstate.inskey = _bt_mkscankey(wstate.index, NULL);
- /* _bt_mkscankey() won't set allequalimage without metapage */
- wstate.inskey->allequalimage = _bt_allequalimage(wstate.index, true);
- wstate.btws_use_wal = RelationNeedsWAL(wstate.index);
+ wstate->heap = btspool->heap;
+ wstate->index = btspool->index;
+ wstate->ub_wstate.smgr_rel = RelationGetSmgr(btspool->index);
+ wstate->ub_wstate.redo = InvalidXLogRecPtr;
+ wstate->inskey = _bt_mkscankey(wstate->index, NULL);
+ /* _bt-mkscankey() won't set allequalimage without metapage */
+ wstate->inskey->allequalimage = _bt_allequalimage(wstate->index, true);
+ wstate->btws_use_wal = RelationNeedsWAL(wstate->index);
/* reserve the metapage */
- wstate.btws_pages_alloced = BTREE_METAPAGE + 1;
- wstate.btws_pages_written = 0;
- wstate.btws_zeropage = NULL; /* until needed */
+ wstate->btws_pages_alloced = 0;
+ wstate->btws_pages_written = 0;
+ wstate->btws_zeropage = NULL;
pgstat_progress_update_param(PROGRESS_CREATEIDX_SUBPHASE,
PROGRESS_BTREE_PHASE_LEAF_LOAD);
- _bt_load(&wstate, btspool, btspool2);
+
+ /*
+ * If not using shared buffers, save the redo pointer location in case a
+ * checkpoint begins during the index build.
+ */
+ if (wstate->_bt_bl_unbuffered_prep && wstate->btws_use_wal)
+ wstate->_bt_bl_unbuffered_prep(&wstate->ub_wstate);
+
+ _bt_load(wstate, btspool, btspool2);
+
+ /*
+ * If not using shared buffers, check if backend must fsync the page
+ * itself.
+ */
+ if (wstate->_bt_bl_unbuffered_finish && wstate->btws_use_wal)
+ wstate->_bt_bl_unbuffered_finish(&wstate->ub_wstate, MAIN_FORKNUM);
}
/*
@@ -609,15 +671,15 @@ _bt_build_callback(Relation index,
}
/*
- * allocate workspace for a new, clean btree page, not linked to any siblings.
+ * Set up workspace for a new, clean btree page, not linked to any siblings.
+ * Caller must allocate the passed in page.
*/
static Page
-_bt_blnewpage(uint32 level)
+_bt_blnewpage(uint32 level, Page page)
{
- Page page;
BTPageOpaque opaque;
- page = (Page) palloc(BLCKSZ);
+ Assert(page);
/* Zero the page and set up standard page header info */
_bt_pageinit(page, BLCKSZ);
@@ -635,18 +697,18 @@ _bt_blnewpage(uint32 level)
return page;
}
-/*
- * emit a completed btree page, and release the working storage.
- */
static void
-_bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
+_bt_blwritepage_direct(BTWriteState *wstate, Page page, BlockNumber blkno, Buffer buf)
{
- /* XLOG stuff */
+ /*
+ * Indexes built outside shared buffers must XLOG the page, issue the page
+ * write request, and take care of fsync'ing the page to the device if a
+ * checkpoint began after the beginning of the index build.
+ *
+ * Use the XLOG_FPI record type for this.
+ */
if (wstate->btws_use_wal)
- {
- /* We use the XLOG_FPI record type for this */
log_newpage(&wstate->index->rd_node, MAIN_FORKNUM, blkno, page, true);
- }
/*
* If we have to write pages nonsequentially, fill in the space with
@@ -660,7 +722,8 @@ _bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
if (!wstate->btws_zeropage)
wstate->btws_zeropage = (Page) palloc0(BLCKSZ);
/* don't set checksum for all-zero page */
- unbuffered_extend(&wstate->ub_wstate, MAIN_FORKNUM, wstate->btws_pages_written++, wstate->btws_zeropage, true);
+ unbuffered_extend(&wstate->ub_wstate, MAIN_FORKNUM,
+ wstate->btws_pages_written++, wstate->btws_zeropage, true);
}
@@ -677,6 +740,61 @@ _bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
pfree(page);
}
+static void
+_bt_blwritepage_shared(BTWriteState *wstate, Page page, BlockNumber blkno, Buffer buf)
+{
+ /*
+ * Indexes built in shared buffers need only to mark the buffer as dirty
+ * and XLOG it.
+ */
+ Assert(buf);
+ START_CRIT_SECTION();
+ MarkBufferDirty(buf);
+ if (wstate->btws_use_wal)
+ log_newpage_buffer(buf, true);
+ END_CRIT_SECTION();
+ _bt_relbuf(wstate->index, buf);
+}
+
+static Page
+_bt_bl_alloc_page_direct(BTWriteState *wstate, BlockNumber *blockno, Buffer *buf)
+{
+ /* buf is only used when using shared buffers, so set it to InvalidBuffer */
+ *buf = InvalidBuffer;
+
+ /*
+ * Assign block number for the page.
+ * This will be used to link to sibling page(s) later and, if this is the
+ * initial page in the level, saved in the BTPageState
+ */
+ *blockno = wstate->btws_pages_alloced++;
+
+ /* now allocate and set up the new page */
+ return palloc(BLCKSZ);
+}
+
+static Page
+_bt_bl_alloc_page_shared(BTWriteState *wstate, BlockNumber *blockno, Buffer *buf)
+{
+ /*
+ * Find a shared buffer for the page. Pass mode RBM_ZERO_AND_LOCK to get an
+ * exclusive lock on the buffer content. No lock on the relation as a whole
+ * is needed (as in LockRelationForExtension()) because the initial index
+ * build is not yet complete.
+ */
+ *buf = ReadBufferExtended(wstate->index, MAIN_FORKNUM, P_NEW,
+ RBM_ZERO_AND_LOCK, NULL);
+
+ /*
+ * bufmgr will assign a block number for the new page.
+ * This will be used to link to sibling page(s) later and, if this is the
+ * initial page in the level, saved in the BTPageState
+ */
+ *blockno = BufferGetBlockNumber(*buf);
+
+ return BufferGetPage(*buf);
+}
+
/*
* allocate and initialize a new BTPageState. the returned structure
* is suitable for immediate use by _bt_buildadd.
@@ -684,13 +802,20 @@ _bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
static BTPageState *
_bt_pagestate(BTWriteState *wstate, uint32 level)
{
+ Buffer buf;
+ BlockNumber blockno;
+
BTPageState *state = (BTPageState *) palloc0(sizeof(BTPageState));
- /* create initial page for level */
- state->btps_page = _bt_blnewpage(level);
+ /*
+ * Allocate and initialize initial page for the level, and if using shared
+ * buffers, extend the relation and allocate a shared buffer for the block.
+ */
+ state->btps_page = _bt_blnewpage(level, wstate->_bt_bl_alloc_page(wstate,
+ &blockno, &buf));
- /* and assign it a page position */
- state->btps_blkno = wstate->btws_pages_alloced++;
+ state->btps_blkno = blockno;
+ state->btps_buf = buf;
state->btps_lowkey = NULL;
/* initialize lastoff so first item goes into P_FIRSTKEY */
@@ -825,6 +950,7 @@ _bt_buildadd(BTWriteState *wstate, BTPageState *state, IndexTuple itup,
{
Page npage;
BlockNumber nblkno;
+ Buffer nbuf;
OffsetNumber last_off;
Size last_truncextra;
Size pgspc;
@@ -839,6 +965,7 @@ _bt_buildadd(BTWriteState *wstate, BTPageState *state, IndexTuple itup,
npage = state->btps_page;
nblkno = state->btps_blkno;
+ nbuf = state->btps_buf;
last_off = state->btps_lastoff;
last_truncextra = state->btps_lastextra;
state->btps_lastextra = truncextra;
@@ -895,15 +1022,14 @@ _bt_buildadd(BTWriteState *wstate, BTPageState *state, IndexTuple itup,
*/
Page opage = npage;
BlockNumber oblkno = nblkno;
+ Buffer obuf = nbuf;
ItemId ii;
ItemId hii;
IndexTuple oitup;
- /* Create new page of same level */
- npage = _bt_blnewpage(state->btps_level);
-
- /* and assign it a page position */
- nblkno = wstate->btws_pages_alloced++;
+ /* Create and initialize a new page of same level */
+ npage = _bt_blnewpage(state->btps_level,
+ wstate->_bt_bl_alloc_page(wstate, &nblkno, &nbuf));
/*
* We copy the last item on the page into the new page, and then
@@ -1013,9 +1139,12 @@ _bt_buildadd(BTWriteState *wstate, BTPageState *state, IndexTuple itup,
/*
* Write out the old page. We never need to touch it again, so we can
- * free the opage workspace too.
+ * free the opage workspace too. obuf has been released and is no longer
+ * valid.
*/
- _bt_blwritepage(wstate, opage, oblkno);
+ wstate->_bt_blwritepage(wstate, opage, oblkno, obuf);
+ obuf = InvalidBuffer;
+ opage = NULL;
/*
* Reset last_off to point to new page
@@ -1050,6 +1179,7 @@ _bt_buildadd(BTWriteState *wstate, BTPageState *state, IndexTuple itup,
state->btps_page = npage;
state->btps_blkno = nblkno;
+ state->btps_buf = nbuf;
state->btps_lastoff = last_off;
}
@@ -1095,12 +1225,12 @@ _bt_sort_dedup_finish_pending(BTWriteState *wstate, BTPageState *state,
* Finish writing out the completed btree.
*/
static void
-_bt_uppershutdown(BTWriteState *wstate, BTPageState *state)
+_bt_uppershutdown(BTWriteState *wstate, BTPageState *state, Buffer metabuf,
+ Page metapage)
{
BTPageState *s;
BlockNumber rootblkno = P_NONE;
uint32 rootlevel = 0;
- Page metapage;
/*
* Each iteration of this loop completes one more level of the tree.
@@ -1146,20 +1276,24 @@ _bt_uppershutdown(BTWriteState *wstate, BTPageState *state)
* back one slot. Then we can dump out the page.
*/
_bt_slideleft(s->btps_page);
- _bt_blwritepage(wstate, s->btps_page, s->btps_blkno);
+ wstate->_bt_blwritepage(wstate, s->btps_page, s->btps_blkno, s->btps_buf);
+ s->btps_buf = InvalidBuffer;
s->btps_page = NULL; /* writepage freed the workspace */
}
/*
- * As the last step in the process, construct the metapage and make it
+ * As the last step in the process, initialize the metapage and make it
* point to the new root (unless we had no data at all, in which case it's
* set to point to "P_NONE"). This changes the index to the "valid" state
* by filling in a valid magic number in the metapage.
+ * After this, metapage will have been freed or invalid and metabuf, if ever
+ * valid, will have been released.
*/
- metapage = (Page) palloc(BLCKSZ);
_bt_initmetapage(metapage, rootblkno, rootlevel,
wstate->inskey->allequalimage);
- _bt_blwritepage(wstate, metapage, BTREE_METAPAGE);
+ wstate->_bt_blwritepage(wstate, metapage, BTREE_METAPAGE, metabuf);
+ metabuf = InvalidBuffer;
+ metapage = NULL;
}
/*
@@ -1169,6 +1303,10 @@ _bt_uppershutdown(BTWriteState *wstate, BTPageState *state)
static void
_bt_load(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
{
+ Page metapage;
+ BlockNumber metablkno;
+ Buffer metabuf;
+
BTPageState *state = NULL;
bool merge = (btspool2 != NULL);
IndexTuple itup,
@@ -1181,11 +1319,30 @@ _bt_load(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
int64 tuples_done = 0;
bool deduplicate;
-
- unbuffered_prep(&wstate->ub_wstate);
deduplicate = wstate->inskey->allequalimage && !btspool->isunique &&
BTGetDeduplicateItems(wstate->index);
+ /*
+ * Reserve block 0 for the metapage up front.
+ *
+ * When using the shared buffers API it is easier to allocate the buffer
+ * for block 0 first instead of trying skip block 0 and allocate it at the
+ * end of index build.
+ *
+ * When not using the shared buffers API, there is no harm in allocating
+ * the metapage first. When block 1 is written, the direct writepage
+ * function will zero-fill block 0. When writing out the metapage at the
+ * end of index build, it will overwrite that block 0.
+ *
+ * The metapage will be initialized and written out at the end of the index
+ * build when all of the information needed to do so is available.
+ *
+ * The block number will always be BTREE_METAPAGE, so the metablkno
+ * variable is unused and only created to avoid a special case in the
+ * direct alloc function.
+ */
+ metapage = wstate->_bt_bl_alloc_page(wstate, &metablkno, &metabuf);
+
if (merge)
{
/*
@@ -1407,10 +1564,7 @@ _bt_load(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
}
/* Close down final pages and write the metapage */
- _bt_uppershutdown(wstate, state);
-
- if (wstate->btws_use_wal)
- unbuffered_finish(&wstate->ub_wstate, MAIN_FORKNUM);
+ _bt_uppershutdown(wstate, state, metabuf, metapage);
}
/*
--
2.30.2
^ permalink raw reply [nested|flat] 152+ messages in thread
* Re: Avoiding smgrimmedsync() during nbtree index builds
2022-01-10 22:50 Re: Avoiding smgrimmedsync() during nbtree index builds Melanie Plageman <[email protected]>
@ 2022-01-11 17:10 ` Melanie Plageman <[email protected]>
2022-01-13 15:52 ` Re: Avoiding smgrimmedsync() during nbtree index builds Justin Pryzby <[email protected]>
2022-01-23 21:55 ` Re: Avoiding smgrimmedsync() during nbtree index builds Andres Freund <[email protected]>
0 siblings, 2 replies; 152+ messages in thread
From: Melanie Plageman @ 2022-01-11 17:10 UTC (permalink / raw)
To: Andres Freund <[email protected]>; +Cc: Heikki Linnakangas <[email protected]>; pgsql-hackers; Peter Geoghegan <[email protected]>; Thomas Munro <[email protected]>
On Mon, Jan 10, 2022 at 5:50 PM Melanie Plageman
<[email protected]> wrote:
>
> I have attached a v3 which includes two commits -- one of which
> implements the directmgr API and uses it and the other which adds
> functionality to use either directmgr or bufmgr API during index build.
>
> Also registering for march commitfest.
Forgot directmgr.h. Attached v4
- Melanie
Attachments:
[text/x-patch] v4-0002-Use-shared-buffers-when-possible-for-index-build.patch (19.4K, ../../CAAKRu_Y735ZMOX5Jz6x5Env=uJhgY1J5CUXtVO5osB5nJ57ypA@mail.gmail.com/2-v4-0002-Use-shared-buffers-when-possible-for-index-build.patch)
download | inline diff:
From 291d5ea92e52f5ac4ff637f383c9bc92973c9e2b Mon Sep 17 00:00:00 2001
From: Melanie Plageman <[email protected]>
Date: Mon, 10 Jan 2022 17:34:01 -0500
Subject: [PATCH v4 2/2] Use shared buffers when possible for index build
When there are not too many tuples, building the index in shared buffers
makes sense. It allows the buffer manager to handle how best to do the
IO.
---
src/backend/access/nbtree/nbtree.c | 34 ++--
src/backend/access/nbtree/nbtsort.c | 268 ++++++++++++++++++++++------
2 files changed, 225 insertions(+), 77 deletions(-)
diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c
index 6ab6651420..0714c7fdd6 100644
--- a/src/backend/access/nbtree/nbtree.c
+++ b/src/backend/access/nbtree/nbtree.c
@@ -151,33 +151,27 @@ bthandler(PG_FUNCTION_ARGS)
void
btbuildempty(Relation index)
{
+ /*
+ * Since this only writes one page, use shared buffers.
+ */
Page metapage;
- UnBufferedWriteState wstate;
-
- wstate.smgr_rel = RelationGetSmgr(index);
-
- unbuffered_prep(&wstate);
-
- /* Construct metapage. */
- metapage = (Page) palloc(BLCKSZ);
- _bt_initmetapage(metapage, P_NONE, 0, _bt_allequalimage(index, false));
+ Buffer metabuf;
/*
- * Write the page and log it. It might seem that an immediate sync would
- * be sufficient to guarantee that the file exists on disk, but recovery
- * itself might remove it while replaying, for example, an
- * XLOG_DBASE_CREATE or XLOG_TBLSPC_CREATE record. Therefore, we need
- * this even when wal_level=minimal.
+ * Allocate a buffer for metapage and initialize metapage.
*/
- unbuffered_write(&wstate, INIT_FORKNUM, BTREE_METAPAGE, metapage);
- log_newpage(&RelationGetSmgr(index)->smgr_rnode.node, INIT_FORKNUM,
- BTREE_METAPAGE, metapage, true);
+ metabuf = ReadBufferExtended(index, INIT_FORKNUM, P_NEW, RBM_ZERO_AND_LOCK, NULL);
+ metapage = BufferGetPage(metabuf);
+ _bt_initmetapage(metapage, P_NONE, 0, _bt_allequalimage(index, false));
/*
- * Even though we xlog'd the page, a concurrent checkpoint may have moved
- * the redo pointer past our xlog record, so we may still need to fsync.
+ * Mark metapage buffer as dirty and XLOG it
*/
- unbuffered_finish(&wstate, INIT_FORKNUM);
+ START_CRIT_SECTION();
+ MarkBufferDirty(metabuf);
+ log_newpage_buffer(metabuf, true);
+ END_CRIT_SECTION();
+ _bt_relbuf(index, metabuf);
}
/*
diff --git a/src/backend/access/nbtree/nbtsort.c b/src/backend/access/nbtree/nbtsort.c
index 5687acd99d..68b53e6c04 100644
--- a/src/backend/access/nbtree/nbtsort.c
+++ b/src/backend/access/nbtree/nbtsort.c
@@ -234,6 +234,7 @@ typedef struct BTPageState
{
Page btps_page; /* workspace for page building */
BlockNumber btps_blkno; /* block # to write this page at */
+ Buffer btps_buf; /* buffer to write this page to */
IndexTuple btps_lowkey; /* page's strict lower bound pivot tuple */
OffsetNumber btps_lastoff; /* last item offset loaded */
Size btps_lastextra; /* last item's extra posting list space */
@@ -251,10 +252,25 @@ typedef struct BTWriteState
Relation index;
BTScanInsert inskey; /* generic insertion scankey */
bool btws_use_wal; /* dump pages to WAL? */
- BlockNumber btws_pages_alloced; /* # pages allocated */
- BlockNumber btws_pages_written; /* # pages written out */
+ BlockNumber btws_pages_alloced; /* # pages allocated for index builds outside SB */
+ BlockNumber btws_pages_written; /* # pages written out for index builds outside SB */
Page btws_zeropage; /* workspace for filling zeroes */
UnBufferedWriteState ub_wstate;
+ /*
+ * Allocate a new btree page. This does not initialize the page.
+ */
+ Page (*_bt_bl_alloc_page) (struct BTWriteState *wstate, BlockNumber
+ *blockno, Buffer *buf);
+ /*
+ * Emit a completed btree page, and release the working storage.
+ */
+ void (*_bt_blwritepage) (struct BTWriteState *wstate, Page page,
+ BlockNumber blkno, Buffer buf);
+
+ void (*_bt_bl_unbuffered_prep) (UnBufferedWriteState *wstate);
+
+ void (*_bt_bl_unbuffered_finish) (UnBufferedWriteState *wstate, ForkNumber
+ forknum);
} BTWriteState;
@@ -263,10 +279,22 @@ static double _bt_spools_heapscan(Relation heap, Relation index,
static void _bt_spooldestroy(BTSpool *btspool);
static void _bt_spool(BTSpool *btspool, ItemPointer self,
Datum *values, bool *isnull);
-static void _bt_leafbuild(BTSpool *btspool, BTSpool *btspool2);
+static void _bt_leafbuild(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2);
static void _bt_build_callback(Relation index, ItemPointer tid, Datum *values,
bool *isnull, bool tupleIsAlive, void *state);
-static Page _bt_blnewpage(uint32 level);
+
+static Page
+_bt_bl_alloc_page_direct(BTWriteState *wstate, BlockNumber *blockno, Buffer *buf);
+static Page
+_bt_bl_alloc_page_shared(BTWriteState *wstate, BlockNumber *blockno, Buffer *buf);
+
+static Page _bt_blnewpage(uint32 level, Page page);
+
+static void
+_bt_blwritepage_direct(BTWriteState *wstate, Page page, BlockNumber blkno, Buffer buf);
+static void
+_bt_blwritepage_shared(BTWriteState *wstate, Page page, BlockNumber blkno, Buffer buf);
+
static BTPageState *_bt_pagestate(BTWriteState *wstate, uint32 level);
static void _bt_slideleft(Page rightmostpage);
static void _bt_sortaddtup(Page page, Size itemsize,
@@ -277,9 +305,10 @@ static void _bt_buildadd(BTWriteState *wstate, BTPageState *state,
static void _bt_sort_dedup_finish_pending(BTWriteState *wstate,
BTPageState *state,
BTDedupState dstate);
-static void _bt_uppershutdown(BTWriteState *wstate, BTPageState *state);
static void _bt_load(BTWriteState *wstate,
BTSpool *btspool, BTSpool *btspool2);
+static void _bt_uppershutdown(BTWriteState *wstate, BTPageState *state, Buffer
+ metabuf, Page metapage);
static void _bt_begin_parallel(BTBuildState *buildstate, bool isconcurrent,
int request);
static void _bt_end_parallel(BTLeader *btleader);
@@ -292,6 +321,21 @@ static void _bt_parallel_scan_and_sort(BTSpool *btspool, BTSpool *btspool2,
Sharedsort *sharedsort2, int sortmem,
bool progress);
+#define BT_BUILD_SB_THRESHOLD 1024
+
+static const BTWriteState wstate_shared = {
+ ._bt_bl_alloc_page = _bt_bl_alloc_page_shared,
+ ._bt_blwritepage = _bt_blwritepage_shared,
+ ._bt_bl_unbuffered_prep = NULL,
+ ._bt_bl_unbuffered_finish = NULL,
+};
+
+static const BTWriteState wstate_direct = {
+ ._bt_bl_alloc_page = _bt_bl_alloc_page_direct,
+ ._bt_blwritepage = _bt_blwritepage_direct,
+ ._bt_bl_unbuffered_prep = unbuffered_prep,
+ ._bt_bl_unbuffered_finish = unbuffered_finish,
+};
/*
* btbuild() -- build a new btree index.
@@ -301,6 +345,7 @@ btbuild(Relation heap, Relation index, IndexInfo *indexInfo)
{
IndexBuildResult *result;
BTBuildState buildstate;
+ BTWriteState writestate;
double reltuples;
#ifdef BTREE_BUILD_STATS
@@ -330,8 +375,12 @@ btbuild(Relation heap, Relation index, IndexInfo *indexInfo)
* Finish the build by (1) completing the sort of the spool file, (2)
* inserting the sorted tuples into btree pages and (3) building the upper
* levels. Finally, it may also be necessary to end use of parallelism.
+ *
+ * Don't use shared buffers if the number of tuples is too large.
*/
- _bt_leafbuild(buildstate.spool, buildstate.spool2);
+ writestate = reltuples < BT_BUILD_SB_THRESHOLD ? wstate_shared : wstate_direct;
+
+ _bt_leafbuild(&writestate, buildstate.spool, buildstate.spool2);
_bt_spooldestroy(buildstate.spool);
if (buildstate.spool2)
_bt_spooldestroy(buildstate.spool2);
@@ -537,10 +586,8 @@ _bt_spool(BTSpool *btspool, ItemPointer self, Datum *values, bool *isnull)
* create an entire btree.
*/
static void
-_bt_leafbuild(BTSpool *btspool, BTSpool *btspool2)
+_bt_leafbuild(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
{
- BTWriteState wstate;
-
#ifdef BTREE_BUILD_STATS
if (log_btree_build_stats)
{
@@ -560,23 +607,38 @@ _bt_leafbuild(BTSpool *btspool, BTSpool *btspool2)
tuplesort_performsort(btspool2->sortstate);
}
- wstate.heap = btspool->heap;
- wstate.index = btspool->index;
- wstate.ub_wstate.smgr_rel = RelationGetSmgr(btspool->index);
- wstate.ub_wstate.redo = InvalidXLogRecPtr;
- wstate.inskey = _bt_mkscankey(wstate.index, NULL);
- /* _bt_mkscankey() won't set allequalimage without metapage */
- wstate.inskey->allequalimage = _bt_allequalimage(wstate.index, true);
- wstate.btws_use_wal = RelationNeedsWAL(wstate.index);
+ wstate->heap = btspool->heap;
+ wstate->index = btspool->index;
+ wstate->ub_wstate.smgr_rel = RelationGetSmgr(btspool->index);
+ wstate->ub_wstate.redo = InvalidXLogRecPtr;
+ wstate->inskey = _bt_mkscankey(wstate->index, NULL);
+ /* _bt-mkscankey() won't set allequalimage without metapage */
+ wstate->inskey->allequalimage = _bt_allequalimage(wstate->index, true);
+ wstate->btws_use_wal = RelationNeedsWAL(wstate->index);
/* reserve the metapage */
- wstate.btws_pages_alloced = BTREE_METAPAGE + 1;
- wstate.btws_pages_written = 0;
- wstate.btws_zeropage = NULL; /* until needed */
+ wstate->btws_pages_alloced = 0;
+ wstate->btws_pages_written = 0;
+ wstate->btws_zeropage = NULL;
pgstat_progress_update_param(PROGRESS_CREATEIDX_SUBPHASE,
PROGRESS_BTREE_PHASE_LEAF_LOAD);
- _bt_load(&wstate, btspool, btspool2);
+
+ /*
+ * If not using shared buffers, save the redo pointer location in case a
+ * checkpoint begins during the index build.
+ */
+ if (wstate->_bt_bl_unbuffered_prep && wstate->btws_use_wal)
+ wstate->_bt_bl_unbuffered_prep(&wstate->ub_wstate);
+
+ _bt_load(wstate, btspool, btspool2);
+
+ /*
+ * If not using shared buffers, check if backend must fsync the page
+ * itself.
+ */
+ if (wstate->_bt_bl_unbuffered_finish && wstate->btws_use_wal)
+ wstate->_bt_bl_unbuffered_finish(&wstate->ub_wstate, MAIN_FORKNUM);
}
/*
@@ -609,15 +671,15 @@ _bt_build_callback(Relation index,
}
/*
- * allocate workspace for a new, clean btree page, not linked to any siblings.
+ * Set up workspace for a new, clean btree page, not linked to any siblings.
+ * Caller must allocate the passed in page.
*/
static Page
-_bt_blnewpage(uint32 level)
+_bt_blnewpage(uint32 level, Page page)
{
- Page page;
BTPageOpaque opaque;
- page = (Page) palloc(BLCKSZ);
+ Assert(page);
/* Zero the page and set up standard page header info */
_bt_pageinit(page, BLCKSZ);
@@ -635,18 +697,18 @@ _bt_blnewpage(uint32 level)
return page;
}
-/*
- * emit a completed btree page, and release the working storage.
- */
static void
-_bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
+_bt_blwritepage_direct(BTWriteState *wstate, Page page, BlockNumber blkno, Buffer buf)
{
- /* XLOG stuff */
+ /*
+ * Indexes built outside shared buffers must XLOG the page, issue the page
+ * write request, and take care of fsync'ing the page to the device if a
+ * checkpoint began after the beginning of the index build.
+ *
+ * Use the XLOG_FPI record type for this.
+ */
if (wstate->btws_use_wal)
- {
- /* We use the XLOG_FPI record type for this */
log_newpage(&wstate->index->rd_node, MAIN_FORKNUM, blkno, page, true);
- }
/*
* If we have to write pages nonsequentially, fill in the space with
@@ -660,7 +722,8 @@ _bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
if (!wstate->btws_zeropage)
wstate->btws_zeropage = (Page) palloc0(BLCKSZ);
/* don't set checksum for all-zero page */
- unbuffered_extend(&wstate->ub_wstate, MAIN_FORKNUM, wstate->btws_pages_written++, wstate->btws_zeropage, true);
+ unbuffered_extend(&wstate->ub_wstate, MAIN_FORKNUM,
+ wstate->btws_pages_written++, wstate->btws_zeropage, true);
}
@@ -677,6 +740,61 @@ _bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
pfree(page);
}
+static void
+_bt_blwritepage_shared(BTWriteState *wstate, Page page, BlockNumber blkno, Buffer buf)
+{
+ /*
+ * Indexes built in shared buffers need only to mark the buffer as dirty
+ * and XLOG it.
+ */
+ Assert(buf);
+ START_CRIT_SECTION();
+ MarkBufferDirty(buf);
+ if (wstate->btws_use_wal)
+ log_newpage_buffer(buf, true);
+ END_CRIT_SECTION();
+ _bt_relbuf(wstate->index, buf);
+}
+
+static Page
+_bt_bl_alloc_page_direct(BTWriteState *wstate, BlockNumber *blockno, Buffer *buf)
+{
+ /* buf is only used when using shared buffers, so set it to InvalidBuffer */
+ *buf = InvalidBuffer;
+
+ /*
+ * Assign block number for the page.
+ * This will be used to link to sibling page(s) later and, if this is the
+ * initial page in the level, saved in the BTPageState
+ */
+ *blockno = wstate->btws_pages_alloced++;
+
+ /* now allocate and set up the new page */
+ return palloc(BLCKSZ);
+}
+
+static Page
+_bt_bl_alloc_page_shared(BTWriteState *wstate, BlockNumber *blockno, Buffer *buf)
+{
+ /*
+ * Find a shared buffer for the page. Pass mode RBM_ZERO_AND_LOCK to get an
+ * exclusive lock on the buffer content. No lock on the relation as a whole
+ * is needed (as in LockRelationForExtension()) because the initial index
+ * build is not yet complete.
+ */
+ *buf = ReadBufferExtended(wstate->index, MAIN_FORKNUM, P_NEW,
+ RBM_ZERO_AND_LOCK, NULL);
+
+ /*
+ * bufmgr will assign a block number for the new page.
+ * This will be used to link to sibling page(s) later and, if this is the
+ * initial page in the level, saved in the BTPageState
+ */
+ *blockno = BufferGetBlockNumber(*buf);
+
+ return BufferGetPage(*buf);
+}
+
/*
* allocate and initialize a new BTPageState. the returned structure
* is suitable for immediate use by _bt_buildadd.
@@ -684,13 +802,20 @@ _bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
static BTPageState *
_bt_pagestate(BTWriteState *wstate, uint32 level)
{
+ Buffer buf;
+ BlockNumber blockno;
+
BTPageState *state = (BTPageState *) palloc0(sizeof(BTPageState));
- /* create initial page for level */
- state->btps_page = _bt_blnewpage(level);
+ /*
+ * Allocate and initialize initial page for the level, and if using shared
+ * buffers, extend the relation and allocate a shared buffer for the block.
+ */
+ state->btps_page = _bt_blnewpage(level, wstate->_bt_bl_alloc_page(wstate,
+ &blockno, &buf));
- /* and assign it a page position */
- state->btps_blkno = wstate->btws_pages_alloced++;
+ state->btps_blkno = blockno;
+ state->btps_buf = buf;
state->btps_lowkey = NULL;
/* initialize lastoff so first item goes into P_FIRSTKEY */
@@ -825,6 +950,7 @@ _bt_buildadd(BTWriteState *wstate, BTPageState *state, IndexTuple itup,
{
Page npage;
BlockNumber nblkno;
+ Buffer nbuf;
OffsetNumber last_off;
Size last_truncextra;
Size pgspc;
@@ -839,6 +965,7 @@ _bt_buildadd(BTWriteState *wstate, BTPageState *state, IndexTuple itup,
npage = state->btps_page;
nblkno = state->btps_blkno;
+ nbuf = state->btps_buf;
last_off = state->btps_lastoff;
last_truncextra = state->btps_lastextra;
state->btps_lastextra = truncextra;
@@ -895,15 +1022,14 @@ _bt_buildadd(BTWriteState *wstate, BTPageState *state, IndexTuple itup,
*/
Page opage = npage;
BlockNumber oblkno = nblkno;
+ Buffer obuf = nbuf;
ItemId ii;
ItemId hii;
IndexTuple oitup;
- /* Create new page of same level */
- npage = _bt_blnewpage(state->btps_level);
-
- /* and assign it a page position */
- nblkno = wstate->btws_pages_alloced++;
+ /* Create and initialize a new page of same level */
+ npage = _bt_blnewpage(state->btps_level,
+ wstate->_bt_bl_alloc_page(wstate, &nblkno, &nbuf));
/*
* We copy the last item on the page into the new page, and then
@@ -1013,9 +1139,12 @@ _bt_buildadd(BTWriteState *wstate, BTPageState *state, IndexTuple itup,
/*
* Write out the old page. We never need to touch it again, so we can
- * free the opage workspace too.
+ * free the opage workspace too. obuf has been released and is no longer
+ * valid.
*/
- _bt_blwritepage(wstate, opage, oblkno);
+ wstate->_bt_blwritepage(wstate, opage, oblkno, obuf);
+ obuf = InvalidBuffer;
+ opage = NULL;
/*
* Reset last_off to point to new page
@@ -1050,6 +1179,7 @@ _bt_buildadd(BTWriteState *wstate, BTPageState *state, IndexTuple itup,
state->btps_page = npage;
state->btps_blkno = nblkno;
+ state->btps_buf = nbuf;
state->btps_lastoff = last_off;
}
@@ -1095,12 +1225,12 @@ _bt_sort_dedup_finish_pending(BTWriteState *wstate, BTPageState *state,
* Finish writing out the completed btree.
*/
static void
-_bt_uppershutdown(BTWriteState *wstate, BTPageState *state)
+_bt_uppershutdown(BTWriteState *wstate, BTPageState *state, Buffer metabuf,
+ Page metapage)
{
BTPageState *s;
BlockNumber rootblkno = P_NONE;
uint32 rootlevel = 0;
- Page metapage;
/*
* Each iteration of this loop completes one more level of the tree.
@@ -1146,20 +1276,24 @@ _bt_uppershutdown(BTWriteState *wstate, BTPageState *state)
* back one slot. Then we can dump out the page.
*/
_bt_slideleft(s->btps_page);
- _bt_blwritepage(wstate, s->btps_page, s->btps_blkno);
+ wstate->_bt_blwritepage(wstate, s->btps_page, s->btps_blkno, s->btps_buf);
+ s->btps_buf = InvalidBuffer;
s->btps_page = NULL; /* writepage freed the workspace */
}
/*
- * As the last step in the process, construct the metapage and make it
+ * As the last step in the process, initialize the metapage and make it
* point to the new root (unless we had no data at all, in which case it's
* set to point to "P_NONE"). This changes the index to the "valid" state
* by filling in a valid magic number in the metapage.
+ * After this, metapage will have been freed or invalid and metabuf, if ever
+ * valid, will have been released.
*/
- metapage = (Page) palloc(BLCKSZ);
_bt_initmetapage(metapage, rootblkno, rootlevel,
wstate->inskey->allequalimage);
- _bt_blwritepage(wstate, metapage, BTREE_METAPAGE);
+ wstate->_bt_blwritepage(wstate, metapage, BTREE_METAPAGE, metabuf);
+ metabuf = InvalidBuffer;
+ metapage = NULL;
}
/*
@@ -1169,6 +1303,10 @@ _bt_uppershutdown(BTWriteState *wstate, BTPageState *state)
static void
_bt_load(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
{
+ Page metapage;
+ BlockNumber metablkno;
+ Buffer metabuf;
+
BTPageState *state = NULL;
bool merge = (btspool2 != NULL);
IndexTuple itup,
@@ -1181,11 +1319,30 @@ _bt_load(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
int64 tuples_done = 0;
bool deduplicate;
-
- unbuffered_prep(&wstate->ub_wstate);
deduplicate = wstate->inskey->allequalimage && !btspool->isunique &&
BTGetDeduplicateItems(wstate->index);
+ /*
+ * Reserve block 0 for the metapage up front.
+ *
+ * When using the shared buffers API it is easier to allocate the buffer
+ * for block 0 first instead of trying skip block 0 and allocate it at the
+ * end of index build.
+ *
+ * When not using the shared buffers API, there is no harm in allocating
+ * the metapage first. When block 1 is written, the direct writepage
+ * function will zero-fill block 0. When writing out the metapage at the
+ * end of index build, it will overwrite that block 0.
+ *
+ * The metapage will be initialized and written out at the end of the index
+ * build when all of the information needed to do so is available.
+ *
+ * The block number will always be BTREE_METAPAGE, so the metablkno
+ * variable is unused and only created to avoid a special case in the
+ * direct alloc function.
+ */
+ metapage = wstate->_bt_bl_alloc_page(wstate, &metablkno, &metabuf);
+
if (merge)
{
/*
@@ -1407,10 +1564,7 @@ _bt_load(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
}
/* Close down final pages and write the metapage */
- _bt_uppershutdown(wstate, state);
-
- if (wstate->btws_use_wal)
- unbuffered_finish(&wstate->ub_wstate, MAIN_FORKNUM);
+ _bt_uppershutdown(wstate, state, metabuf, metapage);
}
/*
--
2.30.2
[text/x-patch] v4-0001-Add-unbuffered-IO-and-avoid-immed-fsync.patch (28.9K, ../../CAAKRu_Y735ZMOX5Jz6x5Env=uJhgY1J5CUXtVO5osB5nJ57ypA@mail.gmail.com/3-v4-0001-Add-unbuffered-IO-and-avoid-immed-fsync.patch)
download | inline diff:
From b016c9ee2e47efe434bf805fb0b310e9a302b0f9 Mon Sep 17 00:00:00 2001
From: Melanie Plageman <[email protected]>
Date: Tue, 28 Sep 2021 14:51:11 -0400
Subject: [PATCH v4 1/2] Add unbuffered IO and avoid immed fsync
Replace unbuffered extends and writes
When writing data outside of shared buffers, the backend must do a
series of steps to ensure the data is both durable and recoverable. When
writing or extending a page of data, the backend must log, checksum, and
write out the page before freeing it.
Additionally, though the data is durable once the WAL entries are on
permanent storage, the XLOG Redo pointer cannot be moved past this WAL
until the page data is safely on permanent storage. If a crash were to
occur before the data is fsync'd, the WAL wouldn't be replayed during
recovery, and the data would be lost. Thus, the backend must ensure that
either the Redo pointer has not moved or that the data is fsync'd before
freeing the page.
This is not a problem with pages written in shared buffers because the
checkpointer will block until all buffers that were dirtied before it
began finish before it moves the Redo pointer past their associated WAL
entries.
This commit makes two main changes:
1) It wraps smgrextend() and smgrwrite() in functions from a new API
for writing data outside of shared buffers, directmgr.
2) It saves the XLOG Redo pointer location before doing the write or
extend. It also adds an fsync request for the page to the
checkpointer's pending-ops table. Then, after doing the write or
extend, if the Redo pointer has moved (meaning a checkpoint has
started since it saved it last), then the backend fsync's the page
itself. Otherwise, it lets the checkpointer take care of fsync'ing
the page the next time it processes the pending-ops table.
---
src/backend/access/gist/gistbuild.c | 17 +++----
src/backend/access/hash/hashpage.c | 9 ++--
src/backend/access/heap/heapam_handler.c | 16 ++++---
src/backend/access/heap/rewriteheap.c | 26 ++++-------
src/backend/access/heap/visibilitymap.c | 8 ++--
src/backend/access/nbtree/nbtree.c | 17 ++++---
src/backend/access/nbtree/nbtsort.c | 39 +++++-----------
src/backend/access/spgist/spginsert.c | 25 +++++-----
src/backend/access/transam/xlog.c | 13 ++++++
src/backend/catalog/storage.c | 25 +++-------
src/backend/storage/Makefile | 2 +-
src/backend/storage/direct/Makefile | 17 +++++++
src/backend/storage/direct/directmgr.c | 57 +++++++++++++++++++++++
src/backend/storage/freespace/freespace.c | 10 ++--
src/include/access/xlog.h | 1 +
src/include/storage/directmgr.h | 44 +++++++++++++++++
16 files changed, 219 insertions(+), 107 deletions(-)
create mode 100644 src/backend/storage/direct/Makefile
create mode 100644 src/backend/storage/direct/directmgr.c
create mode 100644 src/include/storage/directmgr.h
diff --git a/src/backend/access/gist/gistbuild.c b/src/backend/access/gist/gistbuild.c
index 9854116fca..748ca65492 100644
--- a/src/backend/access/gist/gistbuild.c
+++ b/src/backend/access/gist/gistbuild.c
@@ -43,6 +43,7 @@
#include "miscadmin.h"
#include "optimizer/optimizer.h"
#include "storage/bufmgr.h"
+#include "storage/directmgr.h"
#include "storage/smgr.h"
#include "utils/memutils.h"
#include "utils/rel.h"
@@ -91,6 +92,7 @@ typedef struct
int64 indtuples; /* number of tuples indexed */
+ UnBufferedWriteState ub_wstate;
/*
* Extra data structures used during a buffering build. 'gfbb' contains
* information related to managing the build buffers. 'parentMap' is a
@@ -194,6 +196,7 @@ gistbuild(Relation heap, Relation index, IndexInfo *indexInfo)
buildstate.heaprel = heap;
buildstate.sortstate = NULL;
buildstate.giststate = initGISTstate(index);
+ buildstate.ub_wstate.smgr_rel = RelationGetSmgr(index);
/*
* Create a temporary memory context that is reset once for each tuple
@@ -403,14 +406,15 @@ gist_indexsortbuild(GISTBuildState *state)
state->pages_allocated = 0;
state->pages_written = 0;
state->ready_num_pages = 0;
+ unbuffered_prep(&state->ub_wstate);
/*
* Write an empty page as a placeholder for the root page. It will be
* replaced with the real root page at the end.
*/
page = palloc0(BLCKSZ);
- smgrextend(RelationGetSmgr(state->indexrel), MAIN_FORKNUM, GIST_ROOT_BLKNO,
- page, true);
+ unbuffered_extend(&state->ub_wstate, MAIN_FORKNUM, GIST_ROOT_BLKNO, page,
+ true);
state->pages_allocated++;
state->pages_written++;
@@ -450,13 +454,12 @@ gist_indexsortbuild(GISTBuildState *state)
/* Write out the root */
PageSetLSN(pagestate->page, GistBuildLSN);
- PageSetChecksumInplace(pagestate->page, GIST_ROOT_BLKNO);
- smgrwrite(RelationGetSmgr(state->indexrel), MAIN_FORKNUM, GIST_ROOT_BLKNO,
- pagestate->page, true);
+ unbuffered_write(&state->ub_wstate, MAIN_FORKNUM, GIST_ROOT_BLKNO, pagestate->page);
if (RelationNeedsWAL(state->indexrel))
log_newpage(&state->indexrel->rd_node, MAIN_FORKNUM, GIST_ROOT_BLKNO,
pagestate->page, true);
+ unbuffered_finish(&state->ub_wstate, MAIN_FORKNUM);
pfree(pagestate->page);
pfree(pagestate);
}
@@ -570,9 +573,7 @@ gist_indexsortbuild_flush_ready_pages(GISTBuildState *state)
elog(ERROR, "unexpected block number to flush GiST sorting build");
PageSetLSN(page, GistBuildLSN);
- PageSetChecksumInplace(page, blkno);
- smgrextend(RelationGetSmgr(state->indexrel), MAIN_FORKNUM, blkno, page,
- true);
+ unbuffered_extend(&state->ub_wstate, MAIN_FORKNUM, blkno, page, false);
state->pages_written++;
}
diff --git a/src/backend/access/hash/hashpage.c b/src/backend/access/hash/hashpage.c
index ee351aea09..722420adf5 100644
--- a/src/backend/access/hash/hashpage.c
+++ b/src/backend/access/hash/hashpage.c
@@ -32,6 +32,7 @@
#include "access/hash_xlog.h"
#include "miscadmin.h"
#include "port/pg_bitutils.h"
+#include "storage/directmgr.h"
#include "storage/lmgr.h"
#include "storage/predicate.h"
#include "storage/smgr.h"
@@ -990,8 +991,10 @@ _hash_alloc_buckets(Relation rel, BlockNumber firstblock, uint32 nblocks)
PGAlignedBlock zerobuf;
Page page;
HashPageOpaque ovflopaque;
+ UnBufferedWriteState ub_wstate;
lastblock = firstblock + nblocks - 1;
+ ub_wstate.smgr_rel = RelationGetSmgr(rel);
/*
* Check for overflow in block number calculation; if so, we cannot extend
@@ -1000,6 +1003,8 @@ _hash_alloc_buckets(Relation rel, BlockNumber firstblock, uint32 nblocks)
if (lastblock < firstblock || lastblock == InvalidBlockNumber)
return false;
+ unbuffered_prep(&ub_wstate);
+
page = (Page) zerobuf.data;
/*
@@ -1024,9 +1029,7 @@ _hash_alloc_buckets(Relation rel, BlockNumber firstblock, uint32 nblocks)
zerobuf.data,
true);
- PageSetChecksumInplace(page, lastblock);
- smgrextend(RelationGetSmgr(rel), MAIN_FORKNUM, lastblock, zerobuf.data,
- false);
+ unbuffered_extend(&ub_wstate, MAIN_FORKNUM, lastblock, zerobuf.data, false);
return true;
}
diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c
index 39ef8a0b77..8824e39a91 100644
--- a/src/backend/access/heap/heapam_handler.c
+++ b/src/backend/access/heap/heapam_handler.c
@@ -38,6 +38,7 @@
#include "pgstat.h"
#include "storage/bufmgr.h"
#include "storage/bufpage.h"
+#include "storage/directmgr.h"
#include "storage/lmgr.h"
#include "storage/predicate.h"
#include "storage/procarray.h"
@@ -575,6 +576,7 @@ heapam_relation_set_new_filenode(Relation rel,
MultiXactId *minmulti)
{
SMgrRelation srel;
+ UnBufferedWriteState ub_wstate;
/*
* Initialize to the minimum XID that could put tuples in the table. We
@@ -594,15 +596,15 @@ heapam_relation_set_new_filenode(Relation rel,
*minmulti = GetOldestMultiXactId();
srel = RelationCreateStorage(*newrnode, persistence);
+ ub_wstate.smgr_rel = srel;
+ unbuffered_prep(&ub_wstate);
/*
* If required, set up an init fork for an unlogged table so that it can
- * be correctly reinitialized on restart. An immediate sync is required
- * even if the page has been logged, because the write did not go through
- * shared_buffers and therefore a concurrent checkpoint may have moved the
- * redo pointer past our xlog record. Recovery may as well remove it
- * while replaying, for example, XLOG_DBASE_CREATE or XLOG_TBLSPC_CREATE
- * record. Therefore, logging is necessary even if wal_level=minimal.
+ * be correctly reinitialized on restart.
+ * Recovery may as well remove our xlog record while replaying, for
+ * example, XLOG_DBASE_CREATE or XLOG_TBLSPC_CREATE record. Therefore,
+ * logging is necessary even if wal_level=minimal.
*/
if (persistence == RELPERSISTENCE_UNLOGGED)
{
@@ -611,7 +613,7 @@ heapam_relation_set_new_filenode(Relation rel,
rel->rd_rel->relkind == RELKIND_TOASTVALUE);
smgrcreate(srel, INIT_FORKNUM, false);
log_smgrcreate(newrnode, INIT_FORKNUM);
- smgrimmedsync(srel, INIT_FORKNUM);
+ unbuffered_finish(&ub_wstate, INIT_FORKNUM);
}
smgrclose(srel);
diff --git a/src/backend/access/heap/rewriteheap.c b/src/backend/access/heap/rewriteheap.c
index aa265edf60..0bb6eec9f9 100644
--- a/src/backend/access/heap/rewriteheap.c
+++ b/src/backend/access/heap/rewriteheap.c
@@ -119,6 +119,7 @@
#include "replication/logical.h"
#include "replication/slot.h"
#include "storage/bufmgr.h"
+#include "storage/directmgr.h"
#include "storage/fd.h"
#include "storage/procarray.h"
#include "storage/smgr.h"
@@ -152,6 +153,7 @@ typedef struct RewriteStateData
HTAB *rs_old_new_tid_map; /* unmatched B tuples */
HTAB *rs_logical_mappings; /* logical remapping files */
uint32 rs_num_rewrite_mappings; /* # in memory mappings */
+ UnBufferedWriteState rs_unbuffered_wstate;
} RewriteStateData;
/*
@@ -264,6 +266,9 @@ begin_heap_rewrite(Relation old_heap, Relation new_heap, TransactionId oldest_xm
state->rs_freeze_xid = freeze_xid;
state->rs_cutoff_multi = cutoff_multi;
state->rs_cxt = rw_cxt;
+ state->rs_unbuffered_wstate.smgr_rel = RelationGetSmgr(state->rs_new_rel);
+
+ unbuffered_prep(&state->rs_unbuffered_wstate);
/* Initialize hash tables used to track update chains */
hash_ctl.keysize = sizeof(TidHashKey);
@@ -324,21 +329,12 @@ end_heap_rewrite(RewriteState state)
state->rs_buffer,
true);
- PageSetChecksumInplace(state->rs_buffer, state->rs_blockno);
-
- smgrextend(RelationGetSmgr(state->rs_new_rel), MAIN_FORKNUM,
- state->rs_blockno, (char *) state->rs_buffer, true);
+ unbuffered_extend(&state->rs_unbuffered_wstate, MAIN_FORKNUM,
+ state->rs_blockno, state->rs_buffer, false);
}
- /*
- * When we WAL-logged rel pages, we must nonetheless fsync them. The
- * reason is the same as in storage.c's RelationCopyStorage(): we're
- * writing data that's not in shared buffers, and so a CHECKPOINT
- * occurring during the rewriteheap operation won't have fsync'd data we
- * wrote before the checkpoint.
- */
if (RelationNeedsWAL(state->rs_new_rel))
- smgrimmedsync(RelationGetSmgr(state->rs_new_rel), MAIN_FORKNUM);
+ unbuffered_finish(&state->rs_unbuffered_wstate, MAIN_FORKNUM);
logical_end_heap_rewrite(state);
@@ -690,10 +686,8 @@ raw_heap_insert(RewriteState state, HeapTuple tup)
* need for smgr to schedule an fsync for this write; we'll do it
* ourselves in end_heap_rewrite.
*/
- PageSetChecksumInplace(page, state->rs_blockno);
-
- smgrextend(RelationGetSmgr(state->rs_new_rel), MAIN_FORKNUM,
- state->rs_blockno, (char *) page, true);
+ unbuffered_extend(&state->rs_unbuffered_wstate, MAIN_FORKNUM,
+ state->rs_blockno, page, false);
state->rs_blockno++;
state->rs_buffer_valid = false;
diff --git a/src/backend/access/heap/visibilitymap.c b/src/backend/access/heap/visibilitymap.c
index 9032d4758f..1a7482e267 100644
--- a/src/backend/access/heap/visibilitymap.c
+++ b/src/backend/access/heap/visibilitymap.c
@@ -92,6 +92,7 @@
#include "miscadmin.h"
#include "port/pg_bitutils.h"
#include "storage/bufmgr.h"
+#include "storage/directmgr.h"
#include "storage/lmgr.h"
#include "storage/smgr.h"
#include "utils/inval.h"
@@ -616,7 +617,9 @@ vm_extend(Relation rel, BlockNumber vm_nblocks)
BlockNumber vm_nblocks_now;
PGAlignedBlock pg;
SMgrRelation reln;
+ UnBufferedWriteState ub_wstate;
+ ub_wstate.smgr_rel = RelationGetSmgr(rel);
PageInit((Page) pg.data, BLCKSZ, 0);
/*
@@ -654,9 +657,8 @@ vm_extend(Relation rel, BlockNumber vm_nblocks)
/* Now extend the file */
while (vm_nblocks_now < vm_nblocks)
{
- PageSetChecksumInplace((Page) pg.data, vm_nblocks_now);
-
- smgrextend(reln, VISIBILITYMAP_FORKNUM, vm_nblocks_now, pg.data, false);
+ // TODO: aren't these pages empty? why checksum them
+ unbuffered_extend(&ub_wstate, VISIBILITYMAP_FORKNUM, vm_nblocks_now, (Page) pg.data, false);
vm_nblocks_now++;
}
diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c
index 13024af2fa..6ab6651420 100644
--- a/src/backend/access/nbtree/nbtree.c
+++ b/src/backend/access/nbtree/nbtree.c
@@ -29,6 +29,7 @@
#include "pgstat.h"
#include "postmaster/autovacuum.h"
#include "storage/condition_variable.h"
+#include "storage/directmgr.h"
#include "storage/indexfsm.h"
#include "storage/ipc.h"
#include "storage/lmgr.h"
@@ -151,6 +152,11 @@ void
btbuildempty(Relation index)
{
Page metapage;
+ UnBufferedWriteState wstate;
+
+ wstate.smgr_rel = RelationGetSmgr(index);
+
+ unbuffered_prep(&wstate);
/* Construct metapage. */
metapage = (Page) palloc(BLCKSZ);
@@ -163,18 +169,15 @@ btbuildempty(Relation index)
* XLOG_DBASE_CREATE or XLOG_TBLSPC_CREATE record. Therefore, we need
* this even when wal_level=minimal.
*/
- PageSetChecksumInplace(metapage, BTREE_METAPAGE);
- smgrwrite(RelationGetSmgr(index), INIT_FORKNUM, BTREE_METAPAGE,
- (char *) metapage, true);
+ unbuffered_write(&wstate, INIT_FORKNUM, BTREE_METAPAGE, metapage);
log_newpage(&RelationGetSmgr(index)->smgr_rnode.node, INIT_FORKNUM,
BTREE_METAPAGE, metapage, true);
/*
- * An immediate sync is required even if we xlog'd the page, because the
- * write did not go through shared_buffers and therefore a concurrent
- * checkpoint may have moved the redo pointer past our xlog record.
+ * Even though we xlog'd the page, a concurrent checkpoint may have moved
+ * the redo pointer past our xlog record, so we may still need to fsync.
*/
- smgrimmedsync(RelationGetSmgr(index), INIT_FORKNUM);
+ unbuffered_finish(&wstate, INIT_FORKNUM);
}
/*
diff --git a/src/backend/access/nbtree/nbtsort.c b/src/backend/access/nbtree/nbtsort.c
index dc220146fd..5687acd99d 100644
--- a/src/backend/access/nbtree/nbtsort.c
+++ b/src/backend/access/nbtree/nbtsort.c
@@ -57,6 +57,7 @@
#include "executor/instrument.h"
#include "miscadmin.h"
#include "pgstat.h"
+#include "storage/directmgr.h"
#include "storage/smgr.h"
#include "tcop/tcopprot.h" /* pgrminclude ignore */
#include "utils/rel.h"
@@ -253,6 +254,7 @@ typedef struct BTWriteState
BlockNumber btws_pages_alloced; /* # pages allocated */
BlockNumber btws_pages_written; /* # pages written out */
Page btws_zeropage; /* workspace for filling zeroes */
+ UnBufferedWriteState ub_wstate;
} BTWriteState;
@@ -560,6 +562,8 @@ _bt_leafbuild(BTSpool *btspool, BTSpool *btspool2)
wstate.heap = btspool->heap;
wstate.index = btspool->index;
+ wstate.ub_wstate.smgr_rel = RelationGetSmgr(btspool->index);
+ wstate.ub_wstate.redo = InvalidXLogRecPtr;
wstate.inskey = _bt_mkscankey(wstate.index, NULL);
/* _bt_mkscankey() won't set allequalimage without metapage */
wstate.inskey->allequalimage = _bt_allequalimage(wstate.index, true);
@@ -656,31 +660,19 @@ _bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
if (!wstate->btws_zeropage)
wstate->btws_zeropage = (Page) palloc0(BLCKSZ);
/* don't set checksum for all-zero page */
- smgrextend(RelationGetSmgr(wstate->index), MAIN_FORKNUM,
- wstate->btws_pages_written++,
- (char *) wstate->btws_zeropage,
- true);
+ unbuffered_extend(&wstate->ub_wstate, MAIN_FORKNUM, wstate->btws_pages_written++, wstate->btws_zeropage, true);
}
- PageSetChecksumInplace(page, blkno);
- /*
- * Now write the page. There's no need for smgr to schedule an fsync for
- * this write; we'll do it ourselves before ending the build.
- */
+ /* Now write the page. Either we are extending the file... */
if (blkno == wstate->btws_pages_written)
{
- /* extending the file... */
- smgrextend(RelationGetSmgr(wstate->index), MAIN_FORKNUM, blkno,
- (char *) page, true);
+ unbuffered_extend(&wstate->ub_wstate, MAIN_FORKNUM, blkno, page, false);
wstate->btws_pages_written++;
}
+ /* or we are overwriting a block we zero-filled before. */
else
- {
- /* overwriting a block we zero-filled before */
- smgrwrite(RelationGetSmgr(wstate->index), MAIN_FORKNUM, blkno,
- (char *) page, true);
- }
+ unbuffered_write(&wstate->ub_wstate, MAIN_FORKNUM, blkno, page);
pfree(page);
}
@@ -1189,6 +1181,8 @@ _bt_load(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
int64 tuples_done = 0;
bool deduplicate;
+
+ unbuffered_prep(&wstate->ub_wstate);
deduplicate = wstate->inskey->allequalimage && !btspool->isunique &&
BTGetDeduplicateItems(wstate->index);
@@ -1415,17 +1409,8 @@ _bt_load(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
/* Close down final pages and write the metapage */
_bt_uppershutdown(wstate, state);
- /*
- * When we WAL-logged index pages, we must nonetheless fsync index files.
- * Since we're building outside shared buffers, a CHECKPOINT occurring
- * during the build has no way to flush the previously written data to
- * disk (indeed it won't know the index even exists). A crash later on
- * would replay WAL from the checkpoint, therefore it wouldn't replay our
- * earlier WAL entries. If we do not fsync those pages here, they might
- * still not be on disk when the crash occurs.
- */
if (wstate->btws_use_wal)
- smgrimmedsync(RelationGetSmgr(wstate->index), MAIN_FORKNUM);
+ unbuffered_finish(&wstate->ub_wstate, MAIN_FORKNUM);
}
/*
diff --git a/src/backend/access/spgist/spginsert.c b/src/backend/access/spgist/spginsert.c
index bfb74049d0..9fd5686b8d 100644
--- a/src/backend/access/spgist/spginsert.c
+++ b/src/backend/access/spgist/spginsert.c
@@ -25,6 +25,7 @@
#include "catalog/index.h"
#include "miscadmin.h"
#include "storage/bufmgr.h"
+#include "storage/directmgr.h"
#include "storage/smgr.h"
#include "utils/memutils.h"
#include "utils/rel.h"
@@ -156,6 +157,10 @@ void
spgbuildempty(Relation index)
{
Page page;
+ UnBufferedWriteState wstate;
+
+ wstate.smgr_rel = RelationGetSmgr(index);
+ unbuffered_prep(&wstate);
/* Construct metapage. */
page = (Page) palloc(BLCKSZ);
@@ -168,36 +173,30 @@ spgbuildempty(Relation index)
* of their existing content when the corresponding create records are
* replayed.
*/
- PageSetChecksumInplace(page, SPGIST_METAPAGE_BLKNO);
- smgrwrite(RelationGetSmgr(index), INIT_FORKNUM, SPGIST_METAPAGE_BLKNO,
- (char *) page, true);
+ unbuffered_write(&wstate, INIT_FORKNUM, SPGIST_METAPAGE_BLKNO, page);
log_newpage(&(RelationGetSmgr(index))->smgr_rnode.node, INIT_FORKNUM,
SPGIST_METAPAGE_BLKNO, page, true);
/* Likewise for the root page. */
SpGistInitPage(page, SPGIST_LEAF);
- PageSetChecksumInplace(page, SPGIST_ROOT_BLKNO);
- smgrwrite(RelationGetSmgr(index), INIT_FORKNUM, SPGIST_ROOT_BLKNO,
- (char *) page, true);
+ unbuffered_write(&wstate, INIT_FORKNUM, SPGIST_ROOT_BLKNO, page);
log_newpage(&(RelationGetSmgr(index))->smgr_rnode.node, INIT_FORKNUM,
SPGIST_ROOT_BLKNO, page, true);
/* Likewise for the null-tuples root page. */
SpGistInitPage(page, SPGIST_LEAF | SPGIST_NULLS);
- PageSetChecksumInplace(page, SPGIST_NULL_BLKNO);
- smgrwrite(RelationGetSmgr(index), INIT_FORKNUM, SPGIST_NULL_BLKNO,
- (char *) page, true);
+ unbuffered_write(&wstate, INIT_FORKNUM, SPGIST_NULL_BLKNO, page);
log_newpage(&(RelationGetSmgr(index))->smgr_rnode.node, INIT_FORKNUM,
SPGIST_NULL_BLKNO, page, true);
/*
- * An immediate sync is required even if we xlog'd the pages, because the
- * writes did not go through shared buffers and therefore a concurrent
- * checkpoint may have moved the redo pointer past our xlog record.
+ * Because the writes did not go through shared buffers, if a concurrent
+ * checkpoint moved the redo pointer past our xlog record, an immediate
+ * sync is required even if we xlog'd the pages.
*/
- smgrimmedsync(RelationGetSmgr(index), INIT_FORKNUM);
+ unbuffered_finish(&wstate, INIT_FORKNUM);
}
/*
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index c9d4cbf3ff..a06e8d005b 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -13191,6 +13191,19 @@ CheckForStandbyTrigger(void)
return false;
}
+bool RedoRecPtrChanged(XLogRecPtr comparator_ptr)
+{
+ XLogRecPtr ptr;
+ SpinLockAcquire(&XLogCtl->info_lck);
+ ptr = XLogCtl->RedoRecPtr;
+ SpinLockRelease(&XLogCtl->info_lck);
+
+ if (RedoRecPtr < ptr)
+ RedoRecPtr = ptr;
+
+ return RedoRecPtr != comparator_ptr;
+}
+
/*
* Remove the files signaling a standby promotion request.
*/
diff --git a/src/backend/catalog/storage.c b/src/backend/catalog/storage.c
index 9b8075536a..05e66d0434 100644
--- a/src/backend/catalog/storage.c
+++ b/src/backend/catalog/storage.c
@@ -28,6 +28,7 @@
#include "catalog/storage.h"
#include "catalog/storage_xlog.h"
#include "miscadmin.h"
+#include "storage/directmgr.h"
#include "storage/freespace.h"
#include "storage/smgr.h"
#include "utils/hsearch.h"
@@ -420,6 +421,10 @@ RelationCopyStorage(SMgrRelation src, SMgrRelation dst,
bool copying_initfork;
BlockNumber nblocks;
BlockNumber blkno;
+ UnBufferedWriteState wstate;
+
+ wstate.smgr_rel = dst;
+ unbuffered_prep(&wstate);
page = (Page) buf.data;
@@ -477,27 +482,11 @@ RelationCopyStorage(SMgrRelation src, SMgrRelation dst,
if (use_wal)
log_newpage(&dst->smgr_rnode.node, forkNum, blkno, page, false);
- PageSetChecksumInplace(page, blkno);
-
- /*
- * Now write the page. We say skipFsync = true because there's no
- * need for smgr to schedule an fsync for this write; we'll do it
- * ourselves below.
- */
- smgrextend(dst, forkNum, blkno, buf.data, true);
+ unbuffered_extend(&wstate, forkNum, blkno, buf.data, false);
}
- /*
- * When we WAL-logged rel pages, we must nonetheless fsync them. The
- * reason is that since we're copying outside shared buffers, a CHECKPOINT
- * occurring during the copy has no way to flush the previously written
- * data to disk (indeed it won't know the new rel even exists). A crash
- * later on would replay WAL from the checkpoint, therefore it wouldn't
- * replay our earlier WAL entries. If we do not fsync those pages here,
- * they might still not be on disk when the crash occurs.
- */
if (use_wal || copying_initfork)
- smgrimmedsync(dst, forkNum);
+ unbuffered_finish(&wstate, forkNum);
}
/*
diff --git a/src/backend/storage/Makefile b/src/backend/storage/Makefile
index 8376cdfca2..501fae5f9d 100644
--- a/src/backend/storage/Makefile
+++ b/src/backend/storage/Makefile
@@ -8,6 +8,6 @@ subdir = src/backend/storage
top_builddir = ../../..
include $(top_builddir)/src/Makefile.global
-SUBDIRS = buffer file freespace ipc large_object lmgr page smgr sync
+SUBDIRS = buffer direct file freespace ipc large_object lmgr page smgr sync
include $(top_srcdir)/src/backend/common.mk
diff --git a/src/backend/storage/direct/Makefile b/src/backend/storage/direct/Makefile
new file mode 100644
index 0000000000..d82bbed48c
--- /dev/null
+++ b/src/backend/storage/direct/Makefile
@@ -0,0 +1,17 @@
+#-------------------------------------------------------------------------
+#
+# Makefile--
+# Makefile for storage/direct
+#
+# IDENTIFICATION
+# src/backend/storage/direct/Makefile
+#
+#-------------------------------------------------------------------------
+
+subdir = src/backend/storage/direct
+top_builddir = ../../../..
+include $(top_builddir)/src/Makefile.global
+
+OBJS = directmgr.o
+
+include $(top_srcdir)/src/backend/common.mk
diff --git a/src/backend/storage/direct/directmgr.c b/src/backend/storage/direct/directmgr.c
new file mode 100644
index 0000000000..d33c2a7a5f
--- /dev/null
+++ b/src/backend/storage/direct/directmgr.c
@@ -0,0 +1,57 @@
+/*-------------------------------------------------------------------------
+ *
+ * directmgr.c
+ * routines for managing unbuffered IO
+ *
+ * Portions Copyright (c) 1996-2021, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ *
+ * IDENTIFICATION
+ * src/backend/storage/direct/directmgr.c
+ *
+ *-------------------------------------------------------------------------
+ */
+#include "postgres.h"
+
+
+#include "access/xlog.h"
+#include "storage/directmgr.h"
+#include "utils/rel.h"
+
+void unbuffered_prep(UnBufferedWriteState *wstate)
+{
+ wstate->redo = GetRedoRecPtr();
+}
+
+/*
+ * When writing data outside shared buffers, a concurrent CHECKPOINT can move
+ * the redo pointer past our WAL entries and won't flush our data to disk. If
+ * the database crashes before the data makes it to disk, our WAL won't be
+ * replayed and the data will be lost.
+ * Thus, if a CHECKPOINT begins between unbuffered_prep() and
+ * unbuffered_finish(), the backend must fsync the data itself.
+ */
+void unbuffered_finish(UnBufferedWriteState *wstate, ForkNumber forknum)
+{
+ if (RedoRecPtrChanged(wstate->redo))
+ smgrimmedsync(wstate->smgr_rel, forknum);
+}
+
+void
+unbuffered_write(UnBufferedWriteState *wstate, ForkNumber forknum, BlockNumber
+ blocknum, Page page)
+{
+ PageSetChecksumInplace(page, blocknum);
+ smgrwrite(wstate->smgr_rel, forknum, blocknum, (char *) page, false);
+}
+
+void
+unbuffered_extend(UnBufferedWriteState *wstate, ForkNumber forknum, BlockNumber
+ blocknum, Page page, bool empty)
+{
+ if (!empty)
+ PageSetChecksumInplace(page, blocknum);
+ smgrextend(wstate->smgr_rel, forknum, blocknum, (char *) page, false);
+}
+
diff --git a/src/backend/storage/freespace/freespace.c b/src/backend/storage/freespace/freespace.c
index c88cb91f06..d31f75bd60 100644
--- a/src/backend/storage/freespace/freespace.c
+++ b/src/backend/storage/freespace/freespace.c
@@ -26,6 +26,7 @@
#include "access/htup_details.h"
#include "access/xlogutils.h"
#include "miscadmin.h"
+#include "storage/directmgr.h"
#include "storage/freespace.h"
#include "storage/fsm_internals.h"
#include "storage/lmgr.h"
@@ -608,6 +609,9 @@ fsm_extend(Relation rel, BlockNumber fsm_nblocks)
BlockNumber fsm_nblocks_now;
PGAlignedBlock pg;
SMgrRelation reln;
+ UnBufferedWriteState ub_wstate;
+
+ ub_wstate.smgr_rel = RelationGetSmgr(rel);
PageInit((Page) pg.data, BLCKSZ, 0);
@@ -647,10 +651,8 @@ fsm_extend(Relation rel, BlockNumber fsm_nblocks)
/* Extend as needed. */
while (fsm_nblocks_now < fsm_nblocks)
{
- PageSetChecksumInplace((Page) pg.data, fsm_nblocks_now);
-
- smgrextend(reln, FSM_FORKNUM, fsm_nblocks_now,
- pg.data, false);
+ // TODO: why was it checksumming all zero pages?
+ unbuffered_extend(&ub_wstate, FSM_FORKNUM, fsm_nblocks_now, (Page) pg.data, false);
fsm_nblocks_now++;
}
diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h
index bb0c52686a..4961067f81 100644
--- a/src/include/access/xlog.h
+++ b/src/include/access/xlog.h
@@ -312,6 +312,7 @@ extern XLogRecPtr GetInsertRecPtr(void);
extern XLogRecPtr GetFlushRecPtr(TimeLineID *insertTLI);
extern TimeLineID GetWALInsertionTimeLine(void);
extern XLogRecPtr GetLastImportantRecPtr(void);
+extern bool RedoRecPtrChanged(XLogRecPtr comparator_ptr);
extern void RemovePromoteSignalFiles(void);
extern bool PromoteIsTriggered(void);
diff --git a/src/include/storage/directmgr.h b/src/include/storage/directmgr.h
new file mode 100644
index 0000000000..5435ab1d08
--- /dev/null
+++ b/src/include/storage/directmgr.h
@@ -0,0 +1,44 @@
+/*-------------------------------------------------------------------------
+ *
+ * directmgr.h
+ * POSTGRES unbuffered IO manager definitions.
+ *
+ *
+ * Portions Copyright (c) 1996-2021, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * src/include/storage/directmgr.h
+ *
+ *-------------------------------------------------------------------------
+ */
+#include "access/xlogdefs.h"
+#include "common/relpath.h"
+#include "storage/bufpage.h"
+#include "storage/smgr.h"
+#include "utils/relcache.h"
+
+#ifndef DIRECTMGR_H
+#define DIRECTMGR_H
+
+/*
+ * After committing the pg_buffer_stats patch, this will contain a pointer to a
+ * PgBufferAccess struct to count the writes and extends done in this way.
+ */
+typedef struct UnBufferedWriteState
+{
+ SMgrRelation smgr_rel;
+ XLogRecPtr redo;
+} UnBufferedWriteState;
+/*
+ * prototypes for functions in directmgr.c
+ */
+extern void unbuffered_prep(UnBufferedWriteState *wstate);
+extern void unbuffered_finish(UnBufferedWriteState *wstate, ForkNumber forknum);
+extern void
+unbuffered_extend(UnBufferedWriteState *wstate, ForkNumber forknum, BlockNumber
+ blocknum, Page page, bool empty);
+extern void
+unbuffered_write(UnBufferedWriteState *wstate, ForkNumber forknum, BlockNumber
+ blocknum, Page page);
+
+#endif /* DIRECTMGR_H */
--
2.30.2
^ permalink raw reply [nested|flat] 152+ messages in thread
* Re: Avoiding smgrimmedsync() during nbtree index builds
2022-01-10 22:50 Re: Avoiding smgrimmedsync() during nbtree index builds Melanie Plageman <[email protected]>
2022-01-11 17:10 ` Re: Avoiding smgrimmedsync() during nbtree index builds Melanie Plageman <[email protected]>
@ 2022-01-13 15:52 ` Justin Pryzby <[email protected]>
1 sibling, 0 replies; 152+ messages in thread
From: Justin Pryzby @ 2022-01-13 15:52 UTC (permalink / raw)
To: Melanie Plageman <[email protected]>; +Cc: Andres Freund <[email protected]>; Heikki Linnakangas <[email protected]>; pgsql-hackers; Peter Geoghegan <[email protected]>; Thomas Munro <[email protected]>
On Tue, Jan 11, 2022 at 12:10:54PM -0500, Melanie Plageman wrote:
> On Mon, Jan 10, 2022 at 5:50 PM Melanie Plageman <[email protected]> wrote:
> >
> > I have attached a v3 which includes two commits -- one of which
> > implements the directmgr API and uses it and the other which adds
> > functionality to use either directmgr or bufmgr API during index build.
> >
> > Also registering for march commitfest.
>
> Forgot directmgr.h. Attached v4
Thanks - I had reconstructed it first ;)
I think the ifndef should be outside the includes:
> +++ b/src/include/storage/directmgr.h
..
> +#include "access/xlogdefs.h"
..
> +#ifndef DIRECTMGR_H
> +#define DIRECTMGR_H
This is failing on windows CI when I use initdb --data-checksums, as attached.
https://cirrus-ci.com/task/5612464120266752
https://api.cirrus-ci.com/v1/artifact/task/5612464120266752/regress_diffs/src/test/regress/regressio...
+++ c:/cirrus/src/test/regress/results/bitmapops.out 2022-01-13 00:47:46.704621200 +0000
..
+ERROR: could not read block 0 in file "base/16384/30310": read only 0 of 8192 bytes
--
Justin
Attachments:
[text/x-diff] 0003-cirrus-run-initdb-with-data-checksums-for-windows.txt (931B, ../../[email protected]/2-0003-cirrus-run-initdb-with-data-checksums-for-windows.txt)
download | inline diff:
From e1a88c25725bc5b34ca9deb1a0b785048c0c5c28 Mon Sep 17 00:00:00 2001
From: Justin Pryzby <[email protected]>
Date: Wed, 12 Jan 2022 22:31:09 -0600
Subject: [PATCH 3/3] cirrus: run initdb with --data-checksums for windows
---
.cirrus.yml | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/.cirrus.yml b/.cirrus.yml
index 19b3737fa11..efedd1f0873 100644
--- a/.cirrus.yml
+++ b/.cirrus.yml
@@ -390,7 +390,7 @@ task:
- perl src/tools/msvc/vcregress.pl check parallel
startcreate_script:
# paths to binaries need backslashes
- - tmp_install\bin\pg_ctl.exe initdb -D tmp_check/db -l tmp_check/initdb.log
+ - tmp_install\bin\pg_ctl.exe initdb -D tmp_check/db -l tmp_check/initdb.log --options=--data-checksums
- echo include '%TEMP_CONFIG%' >> tmp_check/db/postgresql.conf
- tmp_install\bin\pg_ctl.exe start -D tmp_check/db -l tmp_check/postmaster.log
test_pl_script:
--
2.17.1
^ permalink raw reply [nested|flat] 152+ messages in thread
* Re: Avoiding smgrimmedsync() during nbtree index builds
2022-01-10 22:50 Re: Avoiding smgrimmedsync() during nbtree index builds Melanie Plageman <[email protected]>
2022-01-11 17:10 ` Re: Avoiding smgrimmedsync() during nbtree index builds Melanie Plageman <[email protected]>
@ 2022-01-23 21:55 ` Andres Freund <[email protected]>
2022-02-09 18:49 ` Re: Avoiding smgrimmedsync() during nbtree index builds Melanie Plageman <[email protected]>
1 sibling, 1 reply; 152+ messages in thread
From: Andres Freund @ 2022-01-23 21:55 UTC (permalink / raw)
To: Melanie Plageman <[email protected]>; +Cc: Heikki Linnakangas <[email protected]>; pgsql-hackers; Peter Geoghegan <[email protected]>; Thomas Munro <[email protected]>; Justin Pryzby <[email protected]>
Hi,
On 2022-01-11 12:10:54 -0500, Melanie Plageman wrote:
> On Mon, Jan 10, 2022 at 5:50 PM Melanie Plageman
> <[email protected]> wrote:
> >
> > I have attached a v3 which includes two commits -- one of which
> > implements the directmgr API and uses it and the other which adds
> > functionality to use either directmgr or bufmgr API during index build.
> >
> > Also registering for march commitfest.
>
> Forgot directmgr.h. Attached v4
Are you looking at the failures Justin pointed out? Something isn't quite
right yet. See https://postgr.es/m/20220116202559.GW14051%40telsasoft.com and
the subsequent mail about it also triggering on once on linux.
> Thus, the backend must ensure that
> either the Redo pointer has not moved or that the data is fsync'd before
> freeing the page.
"freeing"?
> This is not a problem with pages written in shared buffers because the
> checkpointer will block until all buffers that were dirtied before it
> began finish before it moves the Redo pointer past their associated WAL
> entries.
> This commit makes two main changes:
>
> 1) It wraps smgrextend() and smgrwrite() in functions from a new API
> for writing data outside of shared buffers, directmgr.
>
> 2) It saves the XLOG Redo pointer location before doing the write or
> extend. It also adds an fsync request for the page to the
> checkpointer's pending-ops table. Then, after doing the write or
> extend, if the Redo pointer has moved (meaning a checkpoint has
> started since it saved it last), then the backend fsync's the page
> itself. Otherwise, it lets the checkpointer take care of fsync'ing
> the page the next time it processes the pending-ops table.
Why combine those two into one commit?
> @@ -654,9 +657,8 @@ vm_extend(Relation rel, BlockNumber vm_nblocks)
> /* Now extend the file */
> while (vm_nblocks_now < vm_nblocks)
> {
> - PageSetChecksumInplace((Page) pg.data, vm_nblocks_now);
> -
> - smgrextend(reln, VISIBILITYMAP_FORKNUM, vm_nblocks_now, pg.data, false);
> + // TODO: aren't these pages empty? why checksum them
> + unbuffered_extend(&ub_wstate, VISIBILITYMAP_FORKNUM, vm_nblocks_now, (Page) pg.data, false);
Yea, it's a bit odd. PageSetChecksumInplace() will just return immediately:
/* If we don't need a checksum, just return */
if (PageIsNew(page) || !DataChecksumsEnabled())
return;
OTOH, it seems easier to have it there than to later forget it, when
e.g. adding some actual initial content to the pages during the smgrextend().
> @@ -560,6 +562,8 @@ _bt_leafbuild(BTSpool *btspool, BTSpool *btspool2)
>
> wstate.heap = btspool->heap;
> wstate.index = btspool->index;
> + wstate.ub_wstate.smgr_rel = RelationGetSmgr(btspool->index);
> + wstate.ub_wstate.redo = InvalidXLogRecPtr;
> wstate.inskey = _bt_mkscankey(wstate.index, NULL);
> /* _bt_mkscankey() won't set allequalimage without metapage */
> wstate.inskey->allequalimage = _bt_allequalimage(wstate.index, true);
> @@ -656,31 +660,19 @@ _bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
> if (!wstate->btws_zeropage)
> wstate->btws_zeropage = (Page) palloc0(BLCKSZ);
> /* don't set checksum for all-zero page */
> - smgrextend(RelationGetSmgr(wstate->index), MAIN_FORKNUM,
> - wstate->btws_pages_written++,
> - (char *) wstate->btws_zeropage,
> - true);
> + unbuffered_extend(&wstate->ub_wstate, MAIN_FORKNUM, wstate->btws_pages_written++, wstate->btws_zeropage, true);
> }
There's a bunch of long lines in here...
> - /*
> - * When we WAL-logged index pages, we must nonetheless fsync index files.
> - * Since we're building outside shared buffers, a CHECKPOINT occurring
> - * during the build has no way to flush the previously written data to
> - * disk (indeed it won't know the index even exists). A crash later on
> - * would replay WAL from the checkpoint, therefore it wouldn't replay our
> - * earlier WAL entries. If we do not fsync those pages here, they might
> - * still not be on disk when the crash occurs.
> - */
> if (wstate->btws_use_wal)
> - smgrimmedsync(RelationGetSmgr(wstate->index), MAIN_FORKNUM);
> + unbuffered_finish(&wstate->ub_wstate, MAIN_FORKNUM);
> }
The API of unbuffered_finish() only sometimes getting called, but
unbuffered_prep() being unconditional, strikes me as prone to bugs. Perhaps
it'd make sense to pass in whether the relation needs to be synced or not instead?
> spgbuildempty(Relation index)
> {
> Page page;
> + UnBufferedWriteState wstate;
> +
> + wstate.smgr_rel = RelationGetSmgr(index);
> + unbuffered_prep(&wstate);
I don't think that's actually safe, and one of the instances could be the
cause cause of the bug CI is seeing:
* Note: since a relcache flush can cause the file handle to be closed again,
* it's unwise to hold onto the pointer returned by this function for any
* long period. Recommended practice is to just re-execute RelationGetSmgr
* each time you need to access the SMgrRelation. It's quite cheap in
* comparison to whatever an smgr function is going to do.
*/
static inline SMgrRelation
RelationGetSmgr(Relation rel)
Greetings,
Andres Freund
^ permalink raw reply [nested|flat] 152+ messages in thread
* Re: Avoiding smgrimmedsync() during nbtree index builds
2022-01-10 22:50 Re: Avoiding smgrimmedsync() during nbtree index builds Melanie Plageman <[email protected]>
2022-01-11 17:10 ` Re: Avoiding smgrimmedsync() during nbtree index builds Melanie Plageman <[email protected]>
2022-01-23 21:55 ` Re: Avoiding smgrimmedsync() during nbtree index builds Andres Freund <[email protected]>
@ 2022-02-09 18:49 ` Melanie Plageman <[email protected]>
2022-02-13 14:33 ` Re: Avoiding smgrimmedsync() during nbtree index builds Dmitry Dolgov <[email protected]>
0 siblings, 1 reply; 152+ messages in thread
From: Melanie Plageman @ 2022-02-09 18:49 UTC (permalink / raw)
To: Andres Freund <[email protected]>; +Cc: Heikki Linnakangas <[email protected]>; pgsql-hackers; Peter Geoghegan <[email protected]>; Thomas Munro <[email protected]>; Justin Pryzby <[email protected]>
Hi,
v5 attached and all email feedback addressed below
On Thu, Jan 13, 2022 at 12:18 PM Robert Haas <[email protected]> wrote:
>
> On Wed, Sep 29, 2021 at 2:36 PM Melanie Plageman
> <[email protected]> wrote:
> > unbuffered_write() and unbuffered_extend() might be able to be used even
> > if unbuffered_prep() and unbuffered_finish() are not used -- for example
> > hash indexes do something I don't entirely understand in which they call
> > smgrextend() directly when allocating buckets but then initialize the
> > new bucket pages using the bufmgr machinery.
>
> My first thought was that someone might do this to make sure that we
> don't run out of disk space after initializing some but not all of the
> buckets. Someone might have some reason for wanting to avoid that
> corner case. However, in _hash_init() that explanation doesn't make
> any sense, because an abort would destroy the entire relation. And in
> _hash_alloc_buckets() the variable "zerobuf" points to a buffer that
> is not, in fact, all zeroes. So my guess is this is just old, crufty
> code - either whatever reasons somebody had for doing it that way are
> no longer valid, or there wasn't any good reason even at the time.
I notice in the comment before _hash_alloc_buckets() is called, it says
/*
* We treat allocation of buckets as a separate WAL-logged action.
* Even if we fail after this operation, won't leak bucket pages;
* rather, the next split will consume this space. In any case, even
* without failure we don't use all the space in one split operation.
*/
Does this mean that it is okay that these pages are written outside of
shared buffers and, though skipFsync is passed as false, a checkpoint
starting and finishing between writing the WAL and
register_dirty_segment() followed by a crash could result in lost data?
On Thu, Jan 13, 2022 at 10:52 AM Justin Pryzby <[email protected]> wrote:
> I think the ifndef should be outside the includes:
Thanks, fixed!
On Sun, Jan 16, 2022 at 3:26 PM Justin Pryzby <[email protected]> wrote:
> Separate from this issue, I wonder if it'd be useful to write a DEBUG log
> showing when btree uses shared_buffers vs fsync. And a regression test which
> first SETs client_min_messages=debug to capture the debug log to demonstrate
> when/that new code path is being hit. I'm not sure if that would be good to
> merge, but it may be useful for now.
I will definitely think about doing this.
On Mon, Jan 17, 2022 at 12:22 PM Justin Pryzby <[email protected]> wrote:
>
> On Sun, Jan 16, 2022 at 02:25:59PM -0600, Justin Pryzby wrote:
> > On Thu, Jan 13, 2022 at 09:52:55AM -0600, Justin Pryzby wrote:
> > > This is failing on windows CI when I use initdb --data-checksums, as attached.
> > >
> > > https://cirrus-ci.com/task/5612464120266752
> > > https://api.cirrus-ci.com/v1/artifact/task/5612464120266752/regress_diffs/src/test/regress/regressio...
> > >
> > > +++ c:/cirrus/src/test/regress/results/bitmapops.out 2022-01-13 00:47:46.704621200 +0000
> > > ..
> > > +ERROR: could not read block 0 in file "base/16384/30310": read only 0 of 8192 bytes
> >
> > The failure isn't consistent, so I double checked my report. I have some more
> > details:
> >
> > The problem occurs maybe only ~25% of the time.
> >
> > The issue is in the 0001 patch.
> >
> > data-checksums isn't necessary to hit the issue.
> >
> > errlocation says: LOCATION: mdread, md.c:686 (the only place the error
> > exists)
> >
> > With Andres' windows crash patch, I obtained a backtrace - attached.
> > https://cirrus-ci.com/task/5978171861368832
> > https://api.cirrus-ci.com/v1/artifact/task/5978171861368832/crashlog/crashlog-postgres.exe_0fa8_2022...
> >
> > Maybe its a race condition or synchronization problem that nowhere else tends
> > to hit.
>
> I meant to say that I had not seen this issue anywhere but windows.
>
> But now, by chance, I still had the 0001 patch in my tree, and hit the same
> issue on linux:
>
> https://cirrus-ci.com/task/4550618281934848
> +++ /tmp/cirrus-ci-build/src/bin/pg_upgrade/tmp_check/regress/results/tuplesort.out 2022-01-17 16:06:35.759108172 +0000
> EXPLAIN (COSTS OFF)
> SELECT id, noabort_increasing, noabort_decreasing FROM abbrev_abort_uuids ORDER BY noabort_increasing LIMIT 5;
> +ERROR: could not read block 0 in file "base/16387/t3_36794": read only 0 of 8192 bytes
Yes, I think this is due to the problem Andres mentioned with my saving
the SMgrRelation and then trying to use it after a relcache flush. The
new patch version addresses this by always re-executing
RelationGetSmgr() as recommended in the comments.
On Sun, Jan 23, 2022 at 4:55 PM Andres Freund <[email protected]> wrote:
> On 2022-01-11 12:10:54 -0500, Melanie Plageman wrote:
> > On Mon, Jan 10, 2022 at 5:50 PM Melanie Plageman
> > <[email protected]> wrote:
> > Thus, the backend must ensure that
> > either the Redo pointer has not moved or that the data is fsync'd before
> > freeing the page.
>
> "freeing"?
Yes, I agree this wording was confusing/incorrect. I meant before it
moves on (I said freeing because it usually pfrees() the page in memory
that it was writing from). I've changed the commit message.
>
> > This is not a problem with pages written in shared buffers because the
> > checkpointer will block until all buffers that were dirtied before it
> > began finish before it moves the Redo pointer past their associated WAL
> > entries.
>
> > This commit makes two main changes:
> >
> > 1) It wraps smgrextend() and smgrwrite() in functions from a new API
> > for writing data outside of shared buffers, directmgr.
> >
> > 2) It saves the XLOG Redo pointer location before doing the write or
> > extend. It also adds an fsync request for the page to the
> > checkpointer's pending-ops table. Then, after doing the write or
> > extend, if the Redo pointer has moved (meaning a checkpoint has
> > started since it saved it last), then the backend fsync's the page
> > itself. Otherwise, it lets the checkpointer take care of fsync'ing
> > the page the next time it processes the pending-ops table.
>
> Why combine those two into one commit?
I've separated it into three commits -- the above two + a separate
commit that actually has the btree index use the self-fsync
optimization.
> > @@ -654,9 +657,8 @@ vm_extend(Relation rel, BlockNumber vm_nblocks)
> > /* Now extend the file */
> > while (vm_nblocks_now < vm_nblocks)
> > {
> > - PageSetChecksumInplace((Page) pg.data, vm_nblocks_now);
> > -
> > - smgrextend(reln, VISIBILITYMAP_FORKNUM, vm_nblocks_now, pg.data, false);
> > + // TODO: aren't these pages empty? why checksum them
> > + unbuffered_extend(&ub_wstate, VISIBILITYMAP_FORKNUM, vm_nblocks_now, (Page) pg.data, false);
>
> Yea, it's a bit odd. PageSetChecksumInplace() will just return immediately:
>
> /* If we don't need a checksum, just return */
> if (PageIsNew(page) || !DataChecksumsEnabled())
> return;
>
> OTOH, it seems easier to have it there than to later forget it, when
> e.g. adding some actual initial content to the pages during the smgrextend().
I've left these as is and removed the comment.
> > @@ -560,6 +562,8 @@ _bt_leafbuild(BTSpool *btspool, BTSpool *btspool2)
> >
> > wstate.heap = btspool->heap;
> > wstate.index = btspool->index;
> > + wstate.ub_wstate.smgr_rel = RelationGetSmgr(btspool->index);
> > + wstate.ub_wstate.redo = InvalidXLogRecPtr;
> > wstate.inskey = _bt_mkscankey(wstate.index, NULL);
> > /* _bt_mkscankey() won't set allequalimage without metapage */
> > wstate.inskey->allequalimage = _bt_allequalimage(wstate.index, true);
> > @@ -656,31 +660,19 @@ _bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
> > if (!wstate->btws_zeropage)
> > wstate->btws_zeropage = (Page) palloc0(BLCKSZ);
> > /* don't set checksum for all-zero page */
> > - smgrextend(RelationGetSmgr(wstate->index), MAIN_FORKNUM,
> > - wstate->btws_pages_written++,
> > - (char *) wstate->btws_zeropage,
> > - true);
> > + unbuffered_extend(&wstate->ub_wstate, MAIN_FORKNUM, wstate->btws_pages_written++, wstate->btws_zeropage, true);
> > }
>
> There's a bunch of long lines in here...
Fixed.
> > - /*
> > - * When we WAL-logged index pages, we must nonetheless fsync index files.
> > - * Since we're building outside shared buffers, a CHECKPOINT occurring
> > - * during the build has no way to flush the previously written data to
> > - * disk (indeed it won't know the index even exists). A crash later on
> > - * would replay WAL from the checkpoint, therefore it wouldn't replay our
> > - * earlier WAL entries. If we do not fsync those pages here, they might
> > - * still not be on disk when the crash occurs.
> > - */
> > if (wstate->btws_use_wal)
> > - smgrimmedsync(RelationGetSmgr(wstate->index), MAIN_FORKNUM);
> > + unbuffered_finish(&wstate->ub_wstate, MAIN_FORKNUM);
> > }
>
> The API of unbuffered_finish() only sometimes getting called, but
> unbuffered_prep() being unconditional, strikes me as prone to bugs. Perhaps
> it'd make sense to pass in whether the relation needs to be synced or not instead?
I've fixed this. Now unbuffered_prep() and unbuffered_finish() will
always be called. I've added a few options to unbuffered_prep() to
indicate whether or not the smgrimmedsync() should be called in the end
as well as whether or not skipFsync should be passed as true or false to
smgrextend() and smgrwrite() and whether or not the avoiding self-fsync
optimization should be used.
I found it best to do it this way because simply passing whether or not
to do the sync to unbuffered_finish() did not allow me to distinguish
between the case in which the sync should not be done ever (because the
caller did not call smgrimmedsync() or because the relation does not
require WAL) and when smgrimmedsync() should only be done if the redo
pointer has changed (in the case of the optimization).
I thought it actually made for a better API to specify up front (in
unbuffered_prep()) whether or not the caller should be prepared to do
the fsync itself or not and whether it not it wanted to do the
optimization. It feels less prone to error and omission.
> > spgbuildempty(Relation index)
> > {
> > Page page;
> > + UnBufferedWriteState wstate;
> > +
> > + wstate.smgr_rel = RelationGetSmgr(index);
> > + unbuffered_prep(&wstate);
>
> I don't think that's actually safe, and one of the instances could be the
> cause cause of the bug CI is seeing:
>
> * Note: since a relcache flush can cause the file handle to be closed again,
> * it's unwise to hold onto the pointer returned by this function for any
> * long period. Recommended practice is to just re-execute RelationGetSmgr
> * each time you need to access the SMgrRelation. It's quite cheap in
> * comparison to whatever an smgr function is going to do.
> */
> static inline SMgrRelation
> RelationGetSmgr(Relation rel)
Yes, I've changed this in the attached v5.
One question I have is whether or not other callers than btree index
could benefit from the self-fsync avoidance optimization.
Also, after taking another look at gist index build, I notice that
smgrimmedsync() is not done anywhere and skipFsync is always passed as
true, so what happens if a full checkpoint and a crash happens between
WAL-logging and whenever the dirty pages make it to permanent storage?
- Melanie
Attachments:
[text/x-patch] v5-0001-Add-unbuffered-IO-API.patch (31.7K, ../../CAAKRu_ZmkCNz-=5U5wZyyi3=Usfs1btsTKc_L9r1ceFvbxE3kg@mail.gmail.com/2-v5-0001-Add-unbuffered-IO-API.patch)
download | inline diff:
From e714a6bd00db76f516ce99bac97ed7e8268eb645 Mon Sep 17 00:00:00 2001
From: Melanie Plageman <[email protected]>
Date: Tue, 8 Feb 2022 19:01:18 -0500
Subject: [PATCH v5 1/4] Add unbuffered IO API
Wrap unbuffered extends and writes in a new API, directmgr.
When writing data outside of shared buffers, the backend must do a
series of steps to ensure the data is both durable and recoverable.
When writing or extending a page of data for a WAL-logged table fork,
the backend must log, checksum (if page is not empty), and write out the
page before moving on.
Additionally, the backend must fsync the page data to ensure it reaches
permanent storage since checkpointer is unaware of the buffer and could
move the Redo pointer past the associated WAL for this write/extend
before it fsyncs the data.
This API is also used for non-WAL-logged and non-self-fsync'd table
forks but with the appropriate exceptions to the above steps.
This commit introduces no functional change. It replaces all current
callers of smgrimmedsync(), smgrextend(), and smgrwrite() with the
equivalent directmgr functions. Consolidating these steps makes IO
outside of shared buffers less error-prone.
---
src/backend/access/gist/gistbuild.c | 36 +++++++----
src/backend/access/hash/hashpage.c | 18 +++---
src/backend/access/heap/heapam_handler.c | 15 +++--
src/backend/access/heap/rewriteheap.c | 53 +++++----------
src/backend/access/heap/visibilitymap.c | 10 ++-
src/backend/access/nbtree/nbtree.c | 18 ++----
src/backend/access/nbtree/nbtsort.c | 56 ++++++----------
src/backend/access/spgist/spginsert.c | 39 ++++-------
src/backend/catalog/storage.c | 30 +++------
src/backend/storage/Makefile | 2 +-
src/backend/storage/direct/Makefile | 17 +++++
src/backend/storage/direct/directmgr.c | 79 +++++++++++++++++++++++
src/backend/storage/freespace/freespace.c | 14 ++--
src/include/catalog/storage.h | 1 +
src/include/storage/directmgr.h | 57 ++++++++++++++++
15 files changed, 276 insertions(+), 169 deletions(-)
create mode 100644 src/backend/storage/direct/Makefile
create mode 100644 src/backend/storage/direct/directmgr.c
create mode 100644 src/include/storage/directmgr.h
diff --git a/src/backend/access/gist/gistbuild.c b/src/backend/access/gist/gistbuild.c
index 4db896a533..42a5e61f8e 100644
--- a/src/backend/access/gist/gistbuild.c
+++ b/src/backend/access/gist/gistbuild.c
@@ -43,6 +43,7 @@
#include "miscadmin.h"
#include "optimizer/optimizer.h"
#include "storage/bufmgr.h"
+#include "storage/directmgr.h"
#include "storage/smgr.h"
#include "utils/memutils.h"
#include "utils/rel.h"
@@ -91,6 +92,7 @@ typedef struct
int64 indtuples; /* number of tuples indexed */
+ UnBufferedWriteState ub_wstate;
/*
* Extra data structures used during a buffering build. 'gfbb' contains
* information related to managing the build buffers. 'parentMap' is a
@@ -409,14 +411,16 @@ gist_indexsortbuild(GISTBuildState *state)
state->pages_allocated = 0;
state->pages_written = 0;
state->ready_num_pages = 0;
+ unbuffered_prep(&state->ub_wstate, false, false);
/*
* Write an empty page as a placeholder for the root page. It will be
* replaced with the real root page at the end.
*/
page = palloc0(BLCKSZ);
- smgrextend(RelationGetSmgr(state->indexrel), MAIN_FORKNUM, GIST_ROOT_BLKNO,
- page, true);
+ unbuffered_extend(&state->ub_wstate, false,
+ RelationGetSmgr(state->indexrel), MAIN_FORKNUM, GIST_ROOT_BLKNO,
+ page, true);
state->pages_allocated++;
state->pages_written++;
@@ -458,12 +462,13 @@ gist_indexsortbuild(GISTBuildState *state)
/* Write out the root */
PageSetLSN(levelstate->pages[0], GistBuildLSN);
- PageSetChecksumInplace(levelstate->pages[0], GIST_ROOT_BLKNO);
- smgrwrite(RelationGetSmgr(state->indexrel), MAIN_FORKNUM, GIST_ROOT_BLKNO,
- levelstate->pages[0], true);
- if (RelationNeedsWAL(state->indexrel))
- log_newpage(&state->indexrel->rd_node, MAIN_FORKNUM, GIST_ROOT_BLKNO,
- levelstate->pages[0], true);
+
+ unbuffered_write(&state->ub_wstate, RelationNeedsWAL(state->indexrel),
+ RelationGetSmgr(state->indexrel), MAIN_FORKNUM, GIST_ROOT_BLKNO,
+ levelstate->pages[0]);
+
+ unbuffered_finish(&state->ub_wstate, RelationGetSmgr(state->indexrel),
+ MAIN_FORKNUM);
pfree(levelstate->pages[0]);
pfree(levelstate);
@@ -633,6 +638,8 @@ gist_indexsortbuild_flush_ready_pages(GISTBuildState *state)
if (state->ready_num_pages == 0)
return;
+ unbuffered_prep(&state->ub_wstate, false, false);
+
for (int i = 0; i < state->ready_num_pages; i++)
{
Page page = state->ready_pages[i];
@@ -643,9 +650,13 @@ gist_indexsortbuild_flush_ready_pages(GISTBuildState *state)
elog(ERROR, "unexpected block number to flush GiST sorting build");
PageSetLSN(page, GistBuildLSN);
- PageSetChecksumInplace(page, blkno);
- smgrextend(RelationGetSmgr(state->indexrel), MAIN_FORKNUM, blkno, page,
- true);
+
+ /*
+ * These will be WAL logged below
+ */
+ unbuffered_extend(&state->ub_wstate, false,
+ RelationGetSmgr(state->indexrel), MAIN_FORKNUM, blkno, page,
+ false);
state->pages_written++;
}
@@ -654,6 +665,9 @@ gist_indexsortbuild_flush_ready_pages(GISTBuildState *state)
log_newpages(&state->indexrel->rd_node, MAIN_FORKNUM, state->ready_num_pages,
state->ready_blknos, state->ready_pages, true);
+ unbuffered_finish(&state->ub_wstate, RelationGetSmgr(state->indexrel),
+ MAIN_FORKNUM);
+
for (int i = 0; i < state->ready_num_pages; i++)
pfree(state->ready_pages[i]);
diff --git a/src/backend/access/hash/hashpage.c b/src/backend/access/hash/hashpage.c
index 28c5297a1d..6096604438 100644
--- a/src/backend/access/hash/hashpage.c
+++ b/src/backend/access/hash/hashpage.c
@@ -33,6 +33,7 @@
#include "access/xloginsert.h"
#include "miscadmin.h"
#include "port/pg_bitutils.h"
+#include "storage/directmgr.h"
#include "storage/lmgr.h"
#include "storage/predicate.h"
#include "storage/smgr.h"
@@ -991,6 +992,7 @@ _hash_alloc_buckets(Relation rel, BlockNumber firstblock, uint32 nblocks)
PGAlignedBlock zerobuf;
Page page;
HashPageOpaque ovflopaque;
+ UnBufferedWriteState ub_wstate;
lastblock = firstblock + nblocks - 1;
@@ -1001,6 +1003,8 @@ _hash_alloc_buckets(Relation rel, BlockNumber firstblock, uint32 nblocks)
if (lastblock < firstblock || lastblock == InvalidBlockNumber)
return false;
+ unbuffered_prep(&ub_wstate, false, true);
+
page = (Page) zerobuf.data;
/*
@@ -1018,16 +1022,10 @@ _hash_alloc_buckets(Relation rel, BlockNumber firstblock, uint32 nblocks)
ovflopaque->hasho_flag = LH_UNUSED_PAGE;
ovflopaque->hasho_page_id = HASHO_PAGE_ID;
- if (RelationNeedsWAL(rel))
- log_newpage(&rel->rd_node,
- MAIN_FORKNUM,
- lastblock,
- zerobuf.data,
- true);
-
- PageSetChecksumInplace(page, lastblock);
- smgrextend(RelationGetSmgr(rel), MAIN_FORKNUM, lastblock, zerobuf.data,
- false);
+ unbuffered_extend(&ub_wstate, RelationNeedsWAL(rel), RelationGetSmgr(rel),
+ MAIN_FORKNUM, lastblock, zerobuf.data, false);
+
+ unbuffered_finish(&ub_wstate, RelationGetSmgr(rel), MAIN_FORKNUM);
return true;
}
diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c
index 39ef8a0b77..9fd6a6f447 100644
--- a/src/backend/access/heap/heapam_handler.c
+++ b/src/backend/access/heap/heapam_handler.c
@@ -38,6 +38,7 @@
#include "pgstat.h"
#include "storage/bufmgr.h"
#include "storage/bufpage.h"
+#include "storage/directmgr.h"
#include "storage/lmgr.h"
#include "storage/predicate.h"
#include "storage/procarray.h"
@@ -575,7 +576,9 @@ heapam_relation_set_new_filenode(Relation rel,
MultiXactId *minmulti)
{
SMgrRelation srel;
+ UnBufferedWriteState ub_wstate;
+ unbuffered_prep(&ub_wstate, true, false);
/*
* Initialize to the minimum XID that could put tuples in the table. We
* know that no xacts older than RecentXmin are still running, so that
@@ -597,12 +600,10 @@ heapam_relation_set_new_filenode(Relation rel,
/*
* If required, set up an init fork for an unlogged table so that it can
- * be correctly reinitialized on restart. An immediate sync is required
- * even if the page has been logged, because the write did not go through
- * shared_buffers and therefore a concurrent checkpoint may have moved the
- * redo pointer past our xlog record. Recovery may as well remove it
- * while replaying, for example, XLOG_DBASE_CREATE or XLOG_TBLSPC_CREATE
- * record. Therefore, logging is necessary even if wal_level=minimal.
+ * be correctly reinitialized on restart.
+ * Recovery may as well remove our xlog record while replaying, for
+ * example, XLOG_DBASE_CREATE or XLOG_TBLSPC_CREATE record. Therefore,
+ * logging is necessary even if wal_level=minimal.
*/
if (persistence == RELPERSISTENCE_UNLOGGED)
{
@@ -611,7 +612,7 @@ heapam_relation_set_new_filenode(Relation rel,
rel->rd_rel->relkind == RELKIND_TOASTVALUE);
smgrcreate(srel, INIT_FORKNUM, false);
log_smgrcreate(newrnode, INIT_FORKNUM);
- smgrimmedsync(srel, INIT_FORKNUM);
+ unbuffered_finish(&ub_wstate, srel, INIT_FORKNUM);
}
smgrclose(srel);
diff --git a/src/backend/access/heap/rewriteheap.c b/src/backend/access/heap/rewriteheap.c
index 2a53826736..12bdd6ff60 100644
--- a/src/backend/access/heap/rewriteheap.c
+++ b/src/backend/access/heap/rewriteheap.c
@@ -119,6 +119,7 @@
#include "replication/logical.h"
#include "replication/slot.h"
#include "storage/bufmgr.h"
+#include "storage/directmgr.h"
#include "storage/fd.h"
#include "storage/procarray.h"
#include "storage/smgr.h"
@@ -152,6 +153,7 @@ typedef struct RewriteStateData
HTAB *rs_old_new_tid_map; /* unmatched B tuples */
HTAB *rs_logical_mappings; /* logical remapping files */
uint32 rs_num_rewrite_mappings; /* # in memory mappings */
+ UnBufferedWriteState rs_unbuffered_wstate;
} RewriteStateData;
/*
@@ -265,6 +267,9 @@ begin_heap_rewrite(Relation old_heap, Relation new_heap, TransactionId oldest_xm
state->rs_cutoff_multi = cutoff_multi;
state->rs_cxt = rw_cxt;
+ unbuffered_prep(&state->rs_unbuffered_wstate,
+ RelationNeedsWAL(state->rs_new_rel), false);
+
/* Initialize hash tables used to track update chains */
hash_ctl.keysize = sizeof(TidHashKey);
hash_ctl.entrysize = sizeof(UnresolvedTupData);
@@ -317,28 +322,14 @@ end_heap_rewrite(RewriteState state)
/* Write the last page, if any */
if (state->rs_buffer_valid)
{
- if (RelationNeedsWAL(state->rs_new_rel))
- log_newpage(&state->rs_new_rel->rd_node,
- MAIN_FORKNUM,
- state->rs_blockno,
- state->rs_buffer,
- true);
-
- PageSetChecksumInplace(state->rs_buffer, state->rs_blockno);
-
- smgrextend(RelationGetSmgr(state->rs_new_rel), MAIN_FORKNUM,
- state->rs_blockno, (char *) state->rs_buffer, true);
+ unbuffered_extend(&state->rs_unbuffered_wstate,
+ RelationNeedsWAL(state->rs_new_rel),
+ RelationGetSmgr(state->rs_new_rel), MAIN_FORKNUM,
+ state->rs_blockno, state->rs_buffer, false);
}
- /*
- * When we WAL-logged rel pages, we must nonetheless fsync them. The
- * reason is the same as in storage.c's RelationCopyStorage(): we're
- * writing data that's not in shared buffers, and so a CHECKPOINT
- * occurring during the rewriteheap operation won't have fsync'd data we
- * wrote before the checkpoint.
- */
- if (RelationNeedsWAL(state->rs_new_rel))
- smgrimmedsync(RelationGetSmgr(state->rs_new_rel), MAIN_FORKNUM);
+ unbuffered_finish(&state->rs_unbuffered_wstate,
+ RelationGetSmgr(state->rs_new_rel), MAIN_FORKNUM);
logical_end_heap_rewrite(state);
@@ -676,24 +667,10 @@ raw_heap_insert(RewriteState state, HeapTuple tup)
* contains a tuple. Hence, unlike RelationGetBufferForTuple(),
* enforce saveFreeSpace unconditionally.
*/
-
- /* XLOG stuff */
- if (RelationNeedsWAL(state->rs_new_rel))
- log_newpage(&state->rs_new_rel->rd_node,
- MAIN_FORKNUM,
- state->rs_blockno,
- page,
- true);
-
- /*
- * Now write the page. We say skipFsync = true because there's no
- * need for smgr to schedule an fsync for this write; we'll do it
- * ourselves in end_heap_rewrite.
- */
- PageSetChecksumInplace(page, state->rs_blockno);
-
- smgrextend(RelationGetSmgr(state->rs_new_rel), MAIN_FORKNUM,
- state->rs_blockno, (char *) page, true);
+ unbuffered_extend(&state->rs_unbuffered_wstate,
+ RelationNeedsWAL(state->rs_new_rel),
+ RelationGetSmgr(state->rs_new_rel), MAIN_FORKNUM,
+ state->rs_blockno, page, false);
state->rs_blockno++;
state->rs_buffer_valid = false;
diff --git a/src/backend/access/heap/visibilitymap.c b/src/backend/access/heap/visibilitymap.c
index e09f25a684..897de5ec1f 100644
--- a/src/backend/access/heap/visibilitymap.c
+++ b/src/backend/access/heap/visibilitymap.c
@@ -93,6 +93,7 @@
#include "miscadmin.h"
#include "port/pg_bitutils.h"
#include "storage/bufmgr.h"
+#include "storage/directmgr.h"
#include "storage/lmgr.h"
#include "storage/smgr.h"
#include "utils/inval.h"
@@ -617,6 +618,7 @@ vm_extend(Relation rel, BlockNumber vm_nblocks)
BlockNumber vm_nblocks_now;
PGAlignedBlock pg;
SMgrRelation reln;
+ UnBufferedWriteState ub_wstate;
PageInit((Page) pg.data, BLCKSZ, 0);
@@ -631,6 +633,7 @@ vm_extend(Relation rel, BlockNumber vm_nblocks)
* by the time we get the lock.
*/
LockRelationForExtension(rel, ExclusiveLock);
+ unbuffered_prep(&ub_wstate, false, true);
/*
* Caution: re-using this smgr pointer could fail if the relcache entry
@@ -655,12 +658,13 @@ vm_extend(Relation rel, BlockNumber vm_nblocks)
/* Now extend the file */
while (vm_nblocks_now < vm_nblocks)
{
- PageSetChecksumInplace((Page) pg.data, vm_nblocks_now);
-
- smgrextend(reln, VISIBILITYMAP_FORKNUM, vm_nblocks_now, pg.data, false);
+ unbuffered_extend(&ub_wstate, false, reln, VISIBILITYMAP_FORKNUM,
+ vm_nblocks_now, (Page) pg.data, false);
vm_nblocks_now++;
}
+ unbuffered_finish(&ub_wstate, reln, VISIBILITYMAP_FORKNUM);
+
/*
* Send a shared-inval message to force other backends to close any smgr
* references they may have for this rel, which we are about to change.
diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c
index c9b4964c1e..1ec7493ad3 100644
--- a/src/backend/access/nbtree/nbtree.c
+++ b/src/backend/access/nbtree/nbtree.c
@@ -30,6 +30,7 @@
#include "pgstat.h"
#include "postmaster/autovacuum.h"
#include "storage/condition_variable.h"
+#include "storage/directmgr.h"
#include "storage/indexfsm.h"
#include "storage/ipc.h"
#include "storage/lmgr.h"
@@ -152,6 +153,9 @@ void
btbuildempty(Relation index)
{
Page metapage;
+ UnBufferedWriteState wstate;
+
+ unbuffered_prep(&wstate, true, false);
/* Construct metapage. */
metapage = (Page) palloc(BLCKSZ);
@@ -164,18 +168,10 @@ btbuildempty(Relation index)
* XLOG_DBASE_CREATE or XLOG_TBLSPC_CREATE record. Therefore, we need
* this even when wal_level=minimal.
*/
- PageSetChecksumInplace(metapage, BTREE_METAPAGE);
- smgrwrite(RelationGetSmgr(index), INIT_FORKNUM, BTREE_METAPAGE,
- (char *) metapage, true);
- log_newpage(&RelationGetSmgr(index)->smgr_rnode.node, INIT_FORKNUM,
- BTREE_METAPAGE, metapage, true);
+ unbuffered_write(&wstate, true, RelationGetSmgr(index), INIT_FORKNUM,
+ BTREE_METAPAGE, metapage);
- /*
- * An immediate sync is required even if we xlog'd the page, because the
- * write did not go through shared_buffers and therefore a concurrent
- * checkpoint may have moved the redo pointer past our xlog record.
- */
- smgrimmedsync(RelationGetSmgr(index), INIT_FORKNUM);
+ unbuffered_finish(&wstate, RelationGetSmgr(index), INIT_FORKNUM);
}
/*
diff --git a/src/backend/access/nbtree/nbtsort.c b/src/backend/access/nbtree/nbtsort.c
index 8a19de2f66..c7a65a9972 100644
--- a/src/backend/access/nbtree/nbtsort.c
+++ b/src/backend/access/nbtree/nbtsort.c
@@ -57,6 +57,7 @@
#include "executor/instrument.h"
#include "miscadmin.h"
#include "pgstat.h"
+#include "storage/directmgr.h"
#include "storage/smgr.h"
#include "tcop/tcopprot.h" /* pgrminclude ignore */
#include "utils/rel.h"
@@ -256,6 +257,7 @@ typedef struct BTWriteState
BlockNumber btws_pages_alloced; /* # pages allocated */
BlockNumber btws_pages_written; /* # pages written out */
Page btws_zeropage; /* workspace for filling zeroes */
+ UnBufferedWriteState ub_wstate;
} BTWriteState;
@@ -643,13 +645,6 @@ _bt_blnewpage(uint32 level)
static void
_bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
{
- /* XLOG stuff */
- if (wstate->btws_use_wal)
- {
- /* We use the XLOG_FPI record type for this */
- log_newpage(&wstate->index->rd_node, MAIN_FORKNUM, blkno, page, true);
- }
-
/*
* If we have to write pages nonsequentially, fill in the space with
* zeroes until we come back and overwrite. This is not logically
@@ -661,32 +656,27 @@ _bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
{
if (!wstate->btws_zeropage)
wstate->btws_zeropage = (Page) palloc0(BLCKSZ);
- /* don't set checksum for all-zero page */
- smgrextend(RelationGetSmgr(wstate->index), MAIN_FORKNUM,
- wstate->btws_pages_written++,
- (char *) wstate->btws_zeropage,
- true);
+
+ unbuffered_extend(&wstate->ub_wstate, false,
+ RelationGetSmgr(wstate->index), MAIN_FORKNUM,
+ wstate->btws_pages_written++, wstate->btws_zeropage, true);
}
- PageSetChecksumInplace(page, blkno);
- /*
- * Now write the page. There's no need for smgr to schedule an fsync for
- * this write; we'll do it ourselves before ending the build.
- */
+ /* Now write the page. Either we are extending the file... */
if (blkno == wstate->btws_pages_written)
{
- /* extending the file... */
- smgrextend(RelationGetSmgr(wstate->index), MAIN_FORKNUM, blkno,
- (char *) page, true);
+ unbuffered_extend(&wstate->ub_wstate, wstate->btws_use_wal,
+ RelationGetSmgr(wstate->index), MAIN_FORKNUM, blkno, page,
+ false);
+
wstate->btws_pages_written++;
}
+
+ /* or we are overwriting a block we zero-filled before. */
else
- {
- /* overwriting a block we zero-filled before */
- smgrwrite(RelationGetSmgr(wstate->index), MAIN_FORKNUM, blkno,
- (char *) page, true);
- }
+ unbuffered_write(&wstate->ub_wstate, wstate->btws_use_wal,
+ RelationGetSmgr(wstate->index), MAIN_FORKNUM, blkno, page);
pfree(page);
}
@@ -1195,6 +1185,9 @@ _bt_load(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
int64 tuples_done = 0;
bool deduplicate;
+
+ unbuffered_prep(&wstate->ub_wstate, wstate->btws_use_wal, false);
+
deduplicate = wstate->inskey->allequalimage && !btspool->isunique &&
BTGetDeduplicateItems(wstate->index);
@@ -1421,17 +1414,8 @@ _bt_load(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
/* Close down final pages and write the metapage */
_bt_uppershutdown(wstate, state);
- /*
- * When we WAL-logged index pages, we must nonetheless fsync index files.
- * Since we're building outside shared buffers, a CHECKPOINT occurring
- * during the build has no way to flush the previously written data to
- * disk (indeed it won't know the index even exists). A crash later on
- * would replay WAL from the checkpoint, therefore it wouldn't replay our
- * earlier WAL entries. If we do not fsync those pages here, they might
- * still not be on disk when the crash occurs.
- */
- if (wstate->btws_use_wal)
- smgrimmedsync(RelationGetSmgr(wstate->index), MAIN_FORKNUM);
+ unbuffered_finish(&wstate->ub_wstate, RelationGetSmgr(wstate->index),
+ MAIN_FORKNUM);
}
/*
diff --git a/src/backend/access/spgist/spginsert.c b/src/backend/access/spgist/spginsert.c
index bfb74049d0..e232ba4b86 100644
--- a/src/backend/access/spgist/spginsert.c
+++ b/src/backend/access/spgist/spginsert.c
@@ -25,6 +25,7 @@
#include "catalog/index.h"
#include "miscadmin.h"
#include "storage/bufmgr.h"
+#include "storage/directmgr.h"
#include "storage/smgr.h"
#include "utils/memutils.h"
#include "utils/rel.h"
@@ -156,48 +157,30 @@ void
spgbuildempty(Relation index)
{
Page page;
+ UnBufferedWriteState wstate;
+
+ unbuffered_prep(&wstate, true, false);
/* Construct metapage. */
page = (Page) palloc(BLCKSZ);
SpGistInitMetapage(page);
- /*
- * Write the page and log it unconditionally. This is important
- * particularly for indexes created on tablespaces and databases whose
- * creation happened after the last redo pointer as recovery removes any
- * of their existing content when the corresponding create records are
- * replayed.
- */
- PageSetChecksumInplace(page, SPGIST_METAPAGE_BLKNO);
- smgrwrite(RelationGetSmgr(index), INIT_FORKNUM, SPGIST_METAPAGE_BLKNO,
- (char *) page, true);
- log_newpage(&(RelationGetSmgr(index))->smgr_rnode.node, INIT_FORKNUM,
- SPGIST_METAPAGE_BLKNO, page, true);
+ unbuffered_write(&wstate, true, RelationGetSmgr(index), INIT_FORKNUM,
+ SPGIST_METAPAGE_BLKNO, page);
/* Likewise for the root page. */
SpGistInitPage(page, SPGIST_LEAF);
- PageSetChecksumInplace(page, SPGIST_ROOT_BLKNO);
- smgrwrite(RelationGetSmgr(index), INIT_FORKNUM, SPGIST_ROOT_BLKNO,
- (char *) page, true);
- log_newpage(&(RelationGetSmgr(index))->smgr_rnode.node, INIT_FORKNUM,
- SPGIST_ROOT_BLKNO, page, true);
+ unbuffered_write(&wstate, true, RelationGetSmgr(index), INIT_FORKNUM,
+ SPGIST_ROOT_BLKNO, page);
/* Likewise for the null-tuples root page. */
SpGistInitPage(page, SPGIST_LEAF | SPGIST_NULLS);
- PageSetChecksumInplace(page, SPGIST_NULL_BLKNO);
- smgrwrite(RelationGetSmgr(index), INIT_FORKNUM, SPGIST_NULL_BLKNO,
- (char *) page, true);
- log_newpage(&(RelationGetSmgr(index))->smgr_rnode.node, INIT_FORKNUM,
- SPGIST_NULL_BLKNO, page, true);
+ unbuffered_write(&wstate, true, RelationGetSmgr(index), INIT_FORKNUM,
+ SPGIST_NULL_BLKNO, page);
- /*
- * An immediate sync is required even if we xlog'd the pages, because the
- * writes did not go through shared buffers and therefore a concurrent
- * checkpoint may have moved the redo pointer past our xlog record.
- */
- smgrimmedsync(RelationGetSmgr(index), INIT_FORKNUM);
+ unbuffered_finish(&wstate, RelationGetSmgr(index), INIT_FORKNUM);
}
/*
diff --git a/src/backend/catalog/storage.c b/src/backend/catalog/storage.c
index 9b8075536a..1ec90e00ab 100644
--- a/src/backend/catalog/storage.c
+++ b/src/backend/catalog/storage.c
@@ -28,6 +28,7 @@
#include "catalog/storage.h"
#include "catalog/storage_xlog.h"
#include "miscadmin.h"
+#include "storage/directmgr.h"
#include "storage/freespace.h"
#include "storage/smgr.h"
#include "utils/hsearch.h"
@@ -420,6 +421,8 @@ RelationCopyStorage(SMgrRelation src, SMgrRelation dst,
bool copying_initfork;
BlockNumber nblocks;
BlockNumber blkno;
+ UnBufferedWriteState wstate;
+
page = (Page) buf.data;
@@ -440,6 +443,8 @@ RelationCopyStorage(SMgrRelation src, SMgrRelation dst,
use_wal = XLogIsNeeded() &&
(relpersistence == RELPERSISTENCE_PERMANENT || copying_initfork);
+ unbuffered_prep(&wstate, (use_wal || copying_initfork), false);
+
nblocks = smgrnblocks(src, forkNum);
for (blkno = 0; blkno < nblocks; blkno++)
@@ -474,30 +479,15 @@ RelationCopyStorage(SMgrRelation src, SMgrRelation dst,
* page this is, so we have to log the full page including any unused
* space.
*/
- if (use_wal)
- log_newpage(&dst->smgr_rnode.node, forkNum, blkno, page, false);
- PageSetChecksumInplace(page, blkno);
+ // TODO: is it okay to pass the page here to unbuffered_extend so that
+ // it can be WAL-logged as a full page even though smgrextend used to
+ // take just buf.data?
+ unbuffered_extend(&wstate, use_wal, dst, forkNum, blkno, page, false);
- /*
- * Now write the page. We say skipFsync = true because there's no
- * need for smgr to schedule an fsync for this write; we'll do it
- * ourselves below.
- */
- smgrextend(dst, forkNum, blkno, buf.data, true);
}
- /*
- * When we WAL-logged rel pages, we must nonetheless fsync them. The
- * reason is that since we're copying outside shared buffers, a CHECKPOINT
- * occurring during the copy has no way to flush the previously written
- * data to disk (indeed it won't know the new rel even exists). A crash
- * later on would replay WAL from the checkpoint, therefore it wouldn't
- * replay our earlier WAL entries. If we do not fsync those pages here,
- * they might still not be on disk when the crash occurs.
- */
- if (use_wal || copying_initfork)
- smgrimmedsync(dst, forkNum);
+ unbuffered_finish(&wstate, dst, forkNum);
}
/*
diff --git a/src/backend/storage/Makefile b/src/backend/storage/Makefile
index 8376cdfca2..501fae5f9d 100644
--- a/src/backend/storage/Makefile
+++ b/src/backend/storage/Makefile
@@ -8,6 +8,6 @@ subdir = src/backend/storage
top_builddir = ../../..
include $(top_builddir)/src/Makefile.global
-SUBDIRS = buffer file freespace ipc large_object lmgr page smgr sync
+SUBDIRS = buffer direct file freespace ipc large_object lmgr page smgr sync
include $(top_srcdir)/src/backend/common.mk
diff --git a/src/backend/storage/direct/Makefile b/src/backend/storage/direct/Makefile
new file mode 100644
index 0000000000..d82bbed48c
--- /dev/null
+++ b/src/backend/storage/direct/Makefile
@@ -0,0 +1,17 @@
+#-------------------------------------------------------------------------
+#
+# Makefile--
+# Makefile for storage/direct
+#
+# IDENTIFICATION
+# src/backend/storage/direct/Makefile
+#
+#-------------------------------------------------------------------------
+
+subdir = src/backend/storage/direct
+top_builddir = ../../../..
+include $(top_builddir)/src/Makefile.global
+
+OBJS = directmgr.o
+
+include $(top_srcdir)/src/backend/common.mk
diff --git a/src/backend/storage/direct/directmgr.c b/src/backend/storage/direct/directmgr.c
new file mode 100644
index 0000000000..371ff5602f
--- /dev/null
+++ b/src/backend/storage/direct/directmgr.c
@@ -0,0 +1,79 @@
+/*-------------------------------------------------------------------------
+ *
+ * directmgr.c
+ * routines for managing unbuffered IO
+ *
+ * Portions Copyright (c) 1996-2021, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ *
+ * IDENTIFICATION
+ * src/backend/storage/direct/directmgr.c
+ *
+ *-------------------------------------------------------------------------
+ */
+#include "postgres.h"
+
+
+#include "access/xloginsert.h"
+#include "storage/directmgr.h"
+
+void
+unbuffered_prep(UnBufferedWriteState *wstate, bool fsync_self, bool
+ request_fsync)
+{
+ wstate->fsync_self = fsync_self;
+ wstate->request_fsync = request_fsync;
+}
+
+void
+unbuffered_write(UnBufferedWriteState *wstate, bool do_wal, SMgrRelation
+ smgrrel, ForkNumber forknum, BlockNumber blocknum, Page page)
+{
+ PageSetChecksumInplace(page, blocknum);
+
+ smgrwrite(smgrrel, forknum, blocknum, (char *) page,
+ !wstate->request_fsync);
+
+ if (do_wal)
+ log_newpage(&(smgrrel)->smgr_rnode.node, forknum,
+ blocknum, page, true);
+}
+
+void
+unbuffered_extend(UnBufferedWriteState *wstate, bool do_wal, SMgrRelation
+ smgrrel, ForkNumber forknum, BlockNumber blocknum, Page page, bool
+ empty)
+{
+ /*
+ * Don't checksum empty pages
+ */
+ if (!empty)
+ PageSetChecksumInplace(page, blocknum);
+
+ smgrextend(smgrrel, forknum, blocknum, (char *) page,
+ !wstate->request_fsync);
+
+ if (do_wal)
+ log_newpage(&(smgrrel)->smgr_rnode.node, forknum,
+ blocknum, page, true);
+
+}
+
+/*
+ * When writing data outside shared buffers, a concurrent CHECKPOINT can move
+ * the redo pointer past our WAL entries and won't flush our data to disk. If
+ * the database crashes before the data makes it to disk, our WAL won't be
+ * replayed and the data will be lost.
+ * Thus, if a CHECKPOINT begins between unbuffered_prep() and
+ * unbuffered_finish(), the backend must fsync the data itself.
+ */
+void
+unbuffered_finish(UnBufferedWriteState *wstate, SMgrRelation smgrrel,
+ ForkNumber forknum)
+{
+ if (!wstate->fsync_self)
+ return;
+
+ smgrimmedsync(smgrrel, forknum);
+}
diff --git a/src/backend/storage/freespace/freespace.c b/src/backend/storage/freespace/freespace.c
index 78c073b7c9..4326ea8f01 100644
--- a/src/backend/storage/freespace/freespace.c
+++ b/src/backend/storage/freespace/freespace.c
@@ -27,6 +27,7 @@
#include "access/xloginsert.h"
#include "access/xlogutils.h"
#include "miscadmin.h"
+#include "storage/directmgr.h"
#include "storage/freespace.h"
#include "storage/fsm_internals.h"
#include "storage/lmgr.h"
@@ -609,9 +610,11 @@ fsm_extend(Relation rel, BlockNumber fsm_nblocks)
BlockNumber fsm_nblocks_now;
PGAlignedBlock pg;
SMgrRelation reln;
+ UnBufferedWriteState ub_wstate;
PageInit((Page) pg.data, BLCKSZ, 0);
+
/*
* We use the relation extension lock to lock out other backends trying to
* extend the FSM at the same time. It also locks out extension of the
@@ -624,6 +627,8 @@ fsm_extend(Relation rel, BlockNumber fsm_nblocks)
*/
LockRelationForExtension(rel, ExclusiveLock);
+ unbuffered_prep(&ub_wstate, false, true);
+
/*
* Caution: re-using this smgr pointer could fail if the relcache entry
* gets closed. It's safe as long as we only do smgr-level operations
@@ -648,14 +653,15 @@ fsm_extend(Relation rel, BlockNumber fsm_nblocks)
/* Extend as needed. */
while (fsm_nblocks_now < fsm_nblocks)
{
- PageSetChecksumInplace((Page) pg.data, fsm_nblocks_now);
-
- smgrextend(reln, FSM_FORKNUM, fsm_nblocks_now,
- pg.data, false);
+ unbuffered_extend(&ub_wstate, false, reln, FSM_FORKNUM,
+ fsm_nblocks_now, (Page) pg.data, false);
fsm_nblocks_now++;
}
+ unbuffered_finish(&ub_wstate, reln, FSM_FORKNUM);
+
UnlockRelationForExtension(rel, ExclusiveLock);
+
}
/*
diff --git a/src/include/catalog/storage.h b/src/include/catalog/storage.h
index 9ffc741913..b097d09c9f 100644
--- a/src/include/catalog/storage.h
+++ b/src/include/catalog/storage.h
@@ -15,6 +15,7 @@
#define STORAGE_H
#include "storage/block.h"
+#include "storage/directmgr.h"
#include "storage/relfilenode.h"
#include "storage/smgr.h"
#include "utils/relcache.h"
diff --git a/src/include/storage/directmgr.h b/src/include/storage/directmgr.h
new file mode 100644
index 0000000000..47653d0d1b
--- /dev/null
+++ b/src/include/storage/directmgr.h
@@ -0,0 +1,57 @@
+/*-------------------------------------------------------------------------
+ *
+ * directmgr.h
+ * POSTGRES unbuffered IO manager definitions.
+ *
+ *
+ * Portions Copyright (c) 1996-2021, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * src/include/storage/directmgr.h
+ *
+ *-------------------------------------------------------------------------
+ */
+#ifndef DIRECTMGR_H
+#define DIRECTMGR_H
+
+#include "common/relpath.h"
+#include "storage/block.h"
+#include "storage/bufpage.h"
+#include "storage/smgr.h"
+
+/*
+ * After committing the pg_buffer_stats patch, this will contain a pointer to a
+ * PgBufferAccess struct to count the writes and extends done in this way.
+ */
+typedef struct UnBufferedWriteState
+{
+ /*
+ * When writing logged table data outside of shared buffers, there is a
+ * risk of a concurrent CHECKPOINT moving the redo pointer past the data's
+ * associated WAL entries. To avoid this, callers in this situation must
+ * fsync the pages they have written themselves.
+ *
+ * Callers able to use the checkpointer's sync request queue when writing
+ * data outside shared buffers (like fsm and vm) can set request_fsync to
+ * true so that these fsync requests are added to the queue.
+ */
+ bool fsync_self;
+ bool request_fsync;
+} UnBufferedWriteState;
+/*
+ * prototypes for functions in directmgr.c
+ */
+extern void
+unbuffered_prep(UnBufferedWriteState *wstate, bool fsync_self, bool
+ request_fsync);
+extern void
+unbuffered_write(UnBufferedWriteState *wstate, bool do_wal, SMgrRelation
+ smgrrel, ForkNumber forknum, BlockNumber blocknum, Page page);
+extern void
+unbuffered_extend(UnBufferedWriteState *wstate, bool do_wal, SMgrRelation smgrrel,
+ ForkNumber forknum, BlockNumber blocknum, Page page, bool empty);
+extern void
+unbuffered_finish(UnBufferedWriteState *wstate, SMgrRelation smgrrel,
+ ForkNumber forknum);
+
+#endif /* DIRECTMGR_H */
--
2.30.2
[text/x-patch] v5-0002-Avoid-immediate-fsync-for-unbuffered-IO.patch (12.2K, ../../CAAKRu_ZmkCNz-=5U5wZyyi3=Usfs1btsTKc_L9r1ceFvbxE3kg@mail.gmail.com/3-v5-0002-Avoid-immediate-fsync-for-unbuffered-IO.patch)
download | inline diff:
From 72c528a913ed4cae1cd11789439bfe1208dd379a Mon Sep 17 00:00:00 2001
From: Melanie Plageman <[email protected]>
Date: Tue, 8 Feb 2022 19:01:27 -0500
Subject: [PATCH v5 2/4] Avoid immediate fsync for unbuffered IO
Data written to WAL-logged table forks is durable once the WAL entries
are on permanent storage; however, the XLOG Redo pointer cannot be moved
past the associated WAL until the page data is safely on permanent
storage. If a crash were to occur before the data is fsync'd, the WAL
wouldn't be replayed during recovery, and the data would be lost.
This is not a problem with pages written in shared buffers because the
checkpointer will block until FlushBuffer() is complete for all buffers
that were dirtied before it began. Therefore it will not move the Redo
pointer past their associated WAL entries until it has fsync'd the data.
A backend writing data outside of shared buffers must ensure that the
data has reached permanent storage itself or that the Redo pointer has
not moved while it was writing the data.
In the common case, the backend should not have to do this fsync itself
and can instead request the checkpointer do it.
To ensure this is safe, the backend can save the XLOG Redo pointer
location before doing the write or extend. Then it can add an fsync
request for the page to the checkpointer's pending-ops table using the
existing mechanism. After doing the write or extend, if the Redo pointer
has moved (meaning a checkpoint has started since it saved it last),
then the backend can simply fsync the page itself. Otherwise, the
checkpointer takes care of fsync'ing the page the next time it processes
the pending-ops table.
This commit adds the optimization option to the directmgr API but does
not add any users, so there is no behavior change.
---
src/backend/access/gist/gistbuild.c | 4 ++--
src/backend/access/hash/hashpage.c | 2 +-
src/backend/access/heap/heapam_handler.c | 2 +-
src/backend/access/heap/rewriteheap.c | 2 +-
src/backend/access/heap/visibilitymap.c | 2 +-
src/backend/access/nbtree/nbtree.c | 2 +-
src/backend/access/nbtree/nbtsort.c | 5 ++++-
src/backend/access/spgist/spginsert.c | 2 +-
src/backend/access/transam/xlog.c | 13 +++++++++++++
src/backend/catalog/storage.c | 2 +-
src/backend/storage/direct/directmgr.c | 18 ++++++++++++++++--
src/backend/storage/freespace/freespace.c | 2 +-
src/include/access/xlog.h | 1 +
src/include/storage/directmgr.h | 13 +++++++++++--
14 files changed, 55 insertions(+), 15 deletions(-)
diff --git a/src/backend/access/gist/gistbuild.c b/src/backend/access/gist/gistbuild.c
index 42a5e61f8e..53226e45bf 100644
--- a/src/backend/access/gist/gistbuild.c
+++ b/src/backend/access/gist/gistbuild.c
@@ -411,7 +411,7 @@ gist_indexsortbuild(GISTBuildState *state)
state->pages_allocated = 0;
state->pages_written = 0;
state->ready_num_pages = 0;
- unbuffered_prep(&state->ub_wstate, false, false);
+ unbuffered_prep(&state->ub_wstate, false, false, false);
/*
* Write an empty page as a placeholder for the root page. It will be
@@ -638,7 +638,7 @@ gist_indexsortbuild_flush_ready_pages(GISTBuildState *state)
if (state->ready_num_pages == 0)
return;
- unbuffered_prep(&state->ub_wstate, false, false);
+ unbuffered_prep(&state->ub_wstate, false, false, false);
for (int i = 0; i < state->ready_num_pages; i++)
{
diff --git a/src/backend/access/hash/hashpage.c b/src/backend/access/hash/hashpage.c
index 6096604438..0c5533e632 100644
--- a/src/backend/access/hash/hashpage.c
+++ b/src/backend/access/hash/hashpage.c
@@ -1003,7 +1003,7 @@ _hash_alloc_buckets(Relation rel, BlockNumber firstblock, uint32 nblocks)
if (lastblock < firstblock || lastblock == InvalidBlockNumber)
return false;
- unbuffered_prep(&ub_wstate, false, true);
+ unbuffered_prep(&ub_wstate, false, false, true);
page = (Page) zerobuf.data;
diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c
index 9fd6a6f447..f9f6527507 100644
--- a/src/backend/access/heap/heapam_handler.c
+++ b/src/backend/access/heap/heapam_handler.c
@@ -578,7 +578,7 @@ heapam_relation_set_new_filenode(Relation rel,
SMgrRelation srel;
UnBufferedWriteState ub_wstate;
- unbuffered_prep(&ub_wstate, true, false);
+ unbuffered_prep(&ub_wstate, false, true, false);
/*
* Initialize to the minimum XID that could put tuples in the table. We
* know that no xacts older than RecentXmin are still running, so that
diff --git a/src/backend/access/heap/rewriteheap.c b/src/backend/access/heap/rewriteheap.c
index 12bdd6ff60..b103a62135 100644
--- a/src/backend/access/heap/rewriteheap.c
+++ b/src/backend/access/heap/rewriteheap.c
@@ -267,7 +267,7 @@ begin_heap_rewrite(Relation old_heap, Relation new_heap, TransactionId oldest_xm
state->rs_cutoff_multi = cutoff_multi;
state->rs_cxt = rw_cxt;
- unbuffered_prep(&state->rs_unbuffered_wstate,
+ unbuffered_prep(&state->rs_unbuffered_wstate, false,
RelationNeedsWAL(state->rs_new_rel), false);
/* Initialize hash tables used to track update chains */
diff --git a/src/backend/access/heap/visibilitymap.c b/src/backend/access/heap/visibilitymap.c
index 897de5ec1f..d844767abc 100644
--- a/src/backend/access/heap/visibilitymap.c
+++ b/src/backend/access/heap/visibilitymap.c
@@ -633,7 +633,7 @@ vm_extend(Relation rel, BlockNumber vm_nblocks)
* by the time we get the lock.
*/
LockRelationForExtension(rel, ExclusiveLock);
- unbuffered_prep(&ub_wstate, false, true);
+ unbuffered_prep(&ub_wstate, false, false, true);
/*
* Caution: re-using this smgr pointer could fail if the relcache entry
diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c
index 1ec7493ad3..843c9e2362 100644
--- a/src/backend/access/nbtree/nbtree.c
+++ b/src/backend/access/nbtree/nbtree.c
@@ -155,7 +155,7 @@ btbuildempty(Relation index)
Page metapage;
UnBufferedWriteState wstate;
- unbuffered_prep(&wstate, true, false);
+ unbuffered_prep(&wstate, false, true, false);
/* Construct metapage. */
metapage = (Page) palloc(BLCKSZ);
diff --git a/src/backend/access/nbtree/nbtsort.c b/src/backend/access/nbtree/nbtsort.c
index c7a65a9972..a67770f3fd 100644
--- a/src/backend/access/nbtree/nbtsort.c
+++ b/src/backend/access/nbtree/nbtsort.c
@@ -1186,7 +1186,10 @@ _bt_load(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
bool deduplicate;
- unbuffered_prep(&wstate->ub_wstate, wstate->btws_use_wal, false);
+ /*
+ * Only bother fsync'ing the data to permanent storage if WAL logging
+ */
+ unbuffered_prep(&wstate->ub_wstate, false, wstate->btws_use_wal, false);
deduplicate = wstate->inskey->allequalimage && !btspool->isunique &&
BTGetDeduplicateItems(wstate->index);
diff --git a/src/backend/access/spgist/spginsert.c b/src/backend/access/spgist/spginsert.c
index e232ba4b86..e45f1f5db9 100644
--- a/src/backend/access/spgist/spginsert.c
+++ b/src/backend/access/spgist/spginsert.c
@@ -159,7 +159,7 @@ spgbuildempty(Relation index)
Page page;
UnBufferedWriteState wstate;
- unbuffered_prep(&wstate, true, false);
+ unbuffered_prep(&wstate, false, true, false);
/* Construct metapage. */
page = (Page) palloc(BLCKSZ);
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 958220c495..c1434d8f85 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -8774,6 +8774,19 @@ GetLastImportantRecPtr(void)
return res;
}
+bool RedoRecPtrChanged(XLogRecPtr comparator_ptr)
+{
+ XLogRecPtr ptr;
+ SpinLockAcquire(&XLogCtl->info_lck);
+ ptr = XLogCtl->RedoRecPtr;
+ SpinLockRelease(&XLogCtl->info_lck);
+
+ if (RedoRecPtr < ptr)
+ RedoRecPtr = ptr;
+
+ return RedoRecPtr != comparator_ptr;
+}
+
/*
* Get the time and LSN of the last xlog segment switch
*/
diff --git a/src/backend/catalog/storage.c b/src/backend/catalog/storage.c
index 1ec90e00ab..307f32ab8c 100644
--- a/src/backend/catalog/storage.c
+++ b/src/backend/catalog/storage.c
@@ -443,7 +443,7 @@ RelationCopyStorage(SMgrRelation src, SMgrRelation dst,
use_wal = XLogIsNeeded() &&
(relpersistence == RELPERSISTENCE_PERMANENT || copying_initfork);
- unbuffered_prep(&wstate, (use_wal || copying_initfork), false);
+ unbuffered_prep(&wstate, false, (use_wal || copying_initfork), false);
nblocks = smgrnblocks(src, forkNum);
diff --git a/src/backend/storage/direct/directmgr.c b/src/backend/storage/direct/directmgr.c
index 371ff5602f..120b7c06a7 100644
--- a/src/backend/storage/direct/directmgr.c
+++ b/src/backend/storage/direct/directmgr.c
@@ -15,15 +15,26 @@
#include "postgres.h"
+#include "access/xlog.h"
#include "access/xloginsert.h"
#include "storage/directmgr.h"
+// TODO: do_optimization can be derived from request_fsync and fsync_self, I
+// think. but is that true in all cases and also is it confusing?
void
-unbuffered_prep(UnBufferedWriteState *wstate, bool fsync_self, bool
- request_fsync)
+unbuffered_prep(UnBufferedWriteState *wstate, bool do_optimization, bool
+ fsync_self, bool request_fsync)
{
+ /*
+ * No reason to do optimization when not required to fsync self
+ */
+ Assert(!do_optimization || (do_optimization && fsync_self));
+
+ wstate->do_optimization = do_optimization;
wstate->fsync_self = fsync_self;
wstate->request_fsync = request_fsync;
+
+ wstate->redo = do_optimization ? GetRedoRecPtr() : InvalidXLogRecPtr;
}
void
@@ -75,5 +86,8 @@ unbuffered_finish(UnBufferedWriteState *wstate, SMgrRelation smgrrel,
if (!wstate->fsync_self)
return;
+ if (wstate->do_optimization && !RedoRecPtrChanged(wstate->redo))
+ return;
+
smgrimmedsync(smgrrel, forknum);
}
diff --git a/src/backend/storage/freespace/freespace.c b/src/backend/storage/freespace/freespace.c
index 4326ea8f01..7c79983ff9 100644
--- a/src/backend/storage/freespace/freespace.c
+++ b/src/backend/storage/freespace/freespace.c
@@ -627,7 +627,7 @@ fsm_extend(Relation rel, BlockNumber fsm_nblocks)
*/
LockRelationForExtension(rel, ExclusiveLock);
- unbuffered_prep(&ub_wstate, false, true);
+ unbuffered_prep(&ub_wstate, false, false, true);
/*
* Caution: re-using this smgr pointer could fail if the relcache entry
diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h
index a4b1c1286f..c2ae0ce304 100644
--- a/src/include/access/xlog.h
+++ b/src/include/access/xlog.h
@@ -310,6 +310,7 @@ extern XLogRecPtr GetInsertRecPtr(void);
extern XLogRecPtr GetFlushRecPtr(TimeLineID *insertTLI);
extern TimeLineID GetWALInsertionTimeLine(void);
extern XLogRecPtr GetLastImportantRecPtr(void);
+extern bool RedoRecPtrChanged(XLogRecPtr comparator_ptr);
extern void RemovePromoteSignalFiles(void);
extern bool PromoteIsTriggered(void);
diff --git a/src/include/storage/directmgr.h b/src/include/storage/directmgr.h
index 47653d0d1b..1ff0c9b0c8 100644
--- a/src/include/storage/directmgr.h
+++ b/src/include/storage/directmgr.h
@@ -14,6 +14,7 @@
#ifndef DIRECTMGR_H
#define DIRECTMGR_H
+#include "access/xlogdefs.h"
#include "common/relpath.h"
#include "storage/block.h"
#include "storage/bufpage.h"
@@ -31,19 +32,27 @@ typedef struct UnBufferedWriteState
* associated WAL entries. To avoid this, callers in this situation must
* fsync the pages they have written themselves.
*
+ * These callers can optionally use the following optimization:
+ * attempt to use the sync request queue and fall back to fsync'ing the
+ * pages themselves if the redo pointer moves between the start and finish
+ * of their write. In order to do this, they must set do_optimization to
+ * true so that the redo pointer is saved before the write begins.
+ *
* Callers able to use the checkpointer's sync request queue when writing
* data outside shared buffers (like fsm and vm) can set request_fsync to
* true so that these fsync requests are added to the queue.
*/
+ bool do_optimization;
bool fsync_self;
bool request_fsync;
+ XLogRecPtr redo;
} UnBufferedWriteState;
/*
* prototypes for functions in directmgr.c
*/
extern void
-unbuffered_prep(UnBufferedWriteState *wstate, bool fsync_self, bool
- request_fsync);
+unbuffered_prep(UnBufferedWriteState *wstate, bool do_optimization, bool
+ fsync_self, bool request_fsync);
extern void
unbuffered_write(UnBufferedWriteState *wstate, bool do_wal, SMgrRelation
smgrrel, ForkNumber forknum, BlockNumber blocknum, Page page);
--
2.30.2
[text/x-patch] v5-0004-Use-shared-buffers-when-possible-for-index-build.patch (19.1K, ../../CAAKRu_ZmkCNz-=5U5wZyyi3=Usfs1btsTKc_L9r1ceFvbxE3kg@mail.gmail.com/4-v5-0004-Use-shared-buffers-when-possible-for-index-build.patch)
download | inline diff:
From 973469e7a5217630542eb7cfd0a7acfc35c6a9fd Mon Sep 17 00:00:00 2001
From: Melanie Plageman <[email protected]>
Date: Tue, 8 Feb 2022 19:01:36 -0500
Subject: [PATCH v5 4/4] Use shared buffers when possible for index build
When there are not too many tuples, building the index in shared buffers
makes sense. It allows the buffer manager to handle how best to do the
IO.
---
src/backend/access/nbtree/nbtree.c | 29 +--
src/backend/access/nbtree/nbtsort.c | 267 ++++++++++++++++++++++------
2 files changed, 223 insertions(+), 73 deletions(-)
diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c
index a1efbe1e6a..4dbf7af9af 100644
--- a/src/backend/access/nbtree/nbtree.c
+++ b/src/backend/access/nbtree/nbtree.c
@@ -152,26 +152,27 @@ bthandler(PG_FUNCTION_ARGS)
void
btbuildempty(Relation index)
{
+ /*
+ * Since this only writes one page, use shared buffers.
+ */
Page metapage;
- UnBufferedWriteState wstate;
-
- unbuffered_prep(&wstate, true, true, true);
+ Buffer metabuf;
- /* Construct metapage. */
- metapage = (Page) palloc(BLCKSZ);
+ /*
+ * Allocate a buffer for metapage and initialize metapage.
+ */
+ metabuf = ReadBufferExtended(index, INIT_FORKNUM, P_NEW, RBM_ZERO_AND_LOCK, NULL);
+ metapage = BufferGetPage(metabuf);
_bt_initmetapage(metapage, P_NONE, 0, _bt_allequalimage(index, false));
/*
- * Write the page and log it. It might seem that an immediate sync would
- * be sufficient to guarantee that the file exists on disk, but recovery
- * itself might remove it while replaying, for example, an
- * XLOG_DBASE_CREATE or XLOG_TBLSPC_CREATE record. Therefore, we need
- * this even when wal_level=minimal.
+ * Mark metapage buffer as dirty and XLOG it
*/
- unbuffered_write(&wstate, true, RelationGetSmgr(index), INIT_FORKNUM,
- BTREE_METAPAGE, metapage);
-
- unbuffered_finish(&wstate, RelationGetSmgr(index), INIT_FORKNUM);
+ START_CRIT_SECTION();
+ MarkBufferDirty(metabuf);
+ log_newpage_buffer(metabuf, true);
+ END_CRIT_SECTION();
+ _bt_relbuf(index, metabuf);
}
/*
diff --git a/src/backend/access/nbtree/nbtsort.c b/src/backend/access/nbtree/nbtsort.c
index 079832bb78..7e1dd93df0 100644
--- a/src/backend/access/nbtree/nbtsort.c
+++ b/src/backend/access/nbtree/nbtsort.c
@@ -237,6 +237,7 @@ typedef struct BTPageState
{
Page btps_page; /* workspace for page building */
BlockNumber btps_blkno; /* block # to write this page at */
+ Buffer btps_buf; /* buffer to write this page to */
IndexTuple btps_lowkey; /* page's strict lower bound pivot tuple */
OffsetNumber btps_lastoff; /* last item offset loaded */
Size btps_lastextra; /* last item's extra posting list space */
@@ -254,10 +255,26 @@ typedef struct BTWriteState
Relation index;
BTScanInsert inskey; /* generic insertion scankey */
bool btws_use_wal; /* dump pages to WAL? */
- BlockNumber btws_pages_alloced; /* # pages allocated */
- BlockNumber btws_pages_written; /* # pages written out */
+ BlockNumber btws_pages_alloced; /* # pages allocated for index builds outside SB */
+ BlockNumber btws_pages_written; /* # pages written out for index builds outside SB */
Page btws_zeropage; /* workspace for filling zeroes */
UnBufferedWriteState ub_wstate;
+ /*
+ * Allocate a new btree page. This does not initialize the page.
+ */
+ Page (*_bt_bl_alloc_page) (struct BTWriteState *wstate, BlockNumber
+ *blockno, Buffer *buf);
+ /*
+ * Emit a completed btree page, and release the working storage.
+ */
+ void (*_bt_blwritepage) (struct BTWriteState *wstate, Page page,
+ BlockNumber blkno, Buffer buf);
+
+ void (*_bt_bl_unbuffered_prep) (UnBufferedWriteState *wstate, bool
+ do_optimization, bool fsync_self, bool request_fsync);
+
+ void (*_bt_bl_unbuffered_finish) (UnBufferedWriteState *wstate,
+ SMgrRelation smgrrel, ForkNumber forknum);
} BTWriteState;
@@ -266,10 +283,22 @@ static double _bt_spools_heapscan(Relation heap, Relation index,
static void _bt_spooldestroy(BTSpool *btspool);
static void _bt_spool(BTSpool *btspool, ItemPointer self,
Datum *values, bool *isnull);
-static void _bt_leafbuild(BTSpool *btspool, BTSpool *btspool2);
+static void _bt_leafbuild(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2);
static void _bt_build_callback(Relation index, ItemPointer tid, Datum *values,
bool *isnull, bool tupleIsAlive, void *state);
-static Page _bt_blnewpage(uint32 level);
+
+static Page
+_bt_bl_alloc_page_direct(BTWriteState *wstate, BlockNumber *blockno, Buffer *buf);
+static Page
+_bt_bl_alloc_page_shared(BTWriteState *wstate, BlockNumber *blockno, Buffer *buf);
+
+static Page _bt_blnewpage(uint32 level, Page page);
+
+static void
+_bt_blwritepage_direct(BTWriteState *wstate, Page page, BlockNumber blkno, Buffer buf);
+static void
+_bt_blwritepage_shared(BTWriteState *wstate, Page page, BlockNumber blkno, Buffer buf);
+
static BTPageState *_bt_pagestate(BTWriteState *wstate, uint32 level);
static void _bt_slideleft(Page rightmostpage);
static void _bt_sortaddtup(Page page, Size itemsize,
@@ -280,9 +309,10 @@ static void _bt_buildadd(BTWriteState *wstate, BTPageState *state,
static void _bt_sort_dedup_finish_pending(BTWriteState *wstate,
BTPageState *state,
BTDedupState dstate);
-static void _bt_uppershutdown(BTWriteState *wstate, BTPageState *state);
static void _bt_load(BTWriteState *wstate,
BTSpool *btspool, BTSpool *btspool2);
+static void _bt_uppershutdown(BTWriteState *wstate, BTPageState *state, Buffer
+ metabuf, Page metapage);
static void _bt_begin_parallel(BTBuildState *buildstate, bool isconcurrent,
int request);
static void _bt_end_parallel(BTLeader *btleader);
@@ -295,6 +325,21 @@ static void _bt_parallel_scan_and_sort(BTSpool *btspool, BTSpool *btspool2,
Sharedsort *sharedsort2, int sortmem,
bool progress);
+#define BT_BUILD_SB_THRESHOLD 1024
+
+static const BTWriteState wstate_shared = {
+ ._bt_bl_alloc_page = _bt_bl_alloc_page_shared,
+ ._bt_blwritepage = _bt_blwritepage_shared,
+ ._bt_bl_unbuffered_prep = NULL,
+ ._bt_bl_unbuffered_finish = NULL,
+};
+
+static const BTWriteState wstate_direct = {
+ ._bt_bl_alloc_page = _bt_bl_alloc_page_direct,
+ ._bt_blwritepage = _bt_blwritepage_direct,
+ ._bt_bl_unbuffered_prep = unbuffered_prep,
+ ._bt_bl_unbuffered_finish = unbuffered_finish,
+};
/*
* btbuild() -- build a new btree index.
@@ -304,6 +349,7 @@ btbuild(Relation heap, Relation index, IndexInfo *indexInfo)
{
IndexBuildResult *result;
BTBuildState buildstate;
+ BTWriteState writestate;
double reltuples;
#ifdef BTREE_BUILD_STATS
@@ -334,8 +380,12 @@ btbuild(Relation heap, Relation index, IndexInfo *indexInfo)
* Finish the build by (1) completing the sort of the spool file, (2)
* inserting the sorted tuples into btree pages and (3) building the upper
* levels. Finally, it may also be necessary to end use of parallelism.
+ *
+ * Don't use shared buffers if the number of tuples is too large.
*/
- _bt_leafbuild(buildstate.spool, buildstate.spool2);
+ writestate = reltuples < BT_BUILD_SB_THRESHOLD ? wstate_shared : wstate_direct;
+
+ _bt_leafbuild(&writestate, buildstate.spool, buildstate.spool2);
_bt_spooldestroy(buildstate.spool);
if (buildstate.spool2)
_bt_spooldestroy(buildstate.spool2);
@@ -543,10 +593,8 @@ _bt_spool(BTSpool *btspool, ItemPointer self, Datum *values, bool *isnull)
* create an entire btree.
*/
static void
-_bt_leafbuild(BTSpool *btspool, BTSpool *btspool2)
+_bt_leafbuild(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
{
- BTWriteState wstate;
-
#ifdef BTREE_BUILD_STATS
if (log_btree_build_stats)
{
@@ -566,21 +614,45 @@ _bt_leafbuild(BTSpool *btspool, BTSpool *btspool2)
tuplesort_performsort(btspool2->sortstate);
}
- wstate.heap = btspool->heap;
- wstate.index = btspool->index;
- wstate.inskey = _bt_mkscankey(wstate.index, NULL);
- /* _bt_mkscankey() won't set allequalimage without metapage */
- wstate.inskey->allequalimage = _bt_allequalimage(wstate.index, true);
- wstate.btws_use_wal = RelationNeedsWAL(wstate.index);
+ wstate->heap = btspool->heap;
+ wstate->index = btspool->index;
+ wstate->inskey = _bt_mkscankey(wstate->index, NULL);
+ /* _bt-mkscankey() won't set allequalimage without metapage */
+ wstate->inskey->allequalimage = _bt_allequalimage(wstate->index, true);
+ wstate->btws_use_wal = RelationNeedsWAL(wstate->index);
/* reserve the metapage */
- wstate.btws_pages_alloced = BTREE_METAPAGE + 1;
- wstate.btws_pages_written = 0;
- wstate.btws_zeropage = NULL; /* until needed */
+ wstate->btws_pages_alloced = 0;
+ wstate->btws_pages_written = 0;
+ wstate->btws_zeropage = NULL;
pgstat_progress_update_param(PROGRESS_CREATEIDX_SUBPHASE,
PROGRESS_BTREE_PHASE_LEAF_LOAD);
- _bt_load(&wstate, btspool, btspool2);
+
+ /*
+ * If not using shared buffers, for a WAL-logged relation, save the redo
+ * pointer location in case a checkpoint begins during the index build.
+ *
+ * This optimization requires that the backend both add an fsync request to
+ * the checkpointer's pending-ops table as well as be prepared to fsync the
+ * page data itself. Because none of these are required if the relation is
+ * not WAL-logged, pass btws_use_wal for all parameters of the prep
+ * function.
+ */
+ if (wstate->_bt_bl_unbuffered_prep)
+ wstate->_bt_bl_unbuffered_prep(&wstate->ub_wstate,
+ wstate->btws_use_wal, wstate->btws_use_wal,
+ wstate->btws_use_wal);
+
+ _bt_load(wstate, btspool, btspool2);
+
+ /*
+ * If not using shared buffers, for a WAL-logged relation, check if backend
+ * must fsync the page itself.
+ */
+ if (wstate->_bt_bl_unbuffered_finish)
+ wstate->_bt_bl_unbuffered_finish(&wstate->ub_wstate,
+ RelationGetSmgr(wstate->index), MAIN_FORKNUM);
}
/*
@@ -613,15 +685,15 @@ _bt_build_callback(Relation index,
}
/*
- * allocate workspace for a new, clean btree page, not linked to any siblings.
+ * Set up workspace for a new, clean btree page, not linked to any siblings.
+ * Caller must allocate the passed in page.
*/
static Page
-_bt_blnewpage(uint32 level)
+_bt_blnewpage(uint32 level, Page page)
{
- Page page;
BTPageOpaque opaque;
- page = (Page) palloc(BLCKSZ);
+ Assert(page);
/* Zero the page and set up standard page header info */
_bt_pageinit(page, BLCKSZ);
@@ -639,11 +711,8 @@ _bt_blnewpage(uint32 level)
return page;
}
-/*
- * emit a completed btree page, and release the working storage.
- */
static void
-_bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
+_bt_blwritepage_direct(BTWriteState *wstate, Page page, BlockNumber blkno, Buffer buf)
{
/*
* If we have to write pages nonsequentially, fill in the space with
@@ -681,6 +750,61 @@ _bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
pfree(page);
}
+static void
+_bt_blwritepage_shared(BTWriteState *wstate, Page page, BlockNumber blkno, Buffer buf)
+{
+ /*
+ * Indexes built in shared buffers need only to mark the buffer as dirty
+ * and XLOG it.
+ */
+ Assert(buf);
+ START_CRIT_SECTION();
+ MarkBufferDirty(buf);
+ if (wstate->btws_use_wal)
+ log_newpage_buffer(buf, true);
+ END_CRIT_SECTION();
+ _bt_relbuf(wstate->index, buf);
+}
+
+static Page
+_bt_bl_alloc_page_direct(BTWriteState *wstate, BlockNumber *blockno, Buffer *buf)
+{
+ /* buf is only used when using shared buffers, so set it to InvalidBuffer */
+ *buf = InvalidBuffer;
+
+ /*
+ * Assign block number for the page.
+ * This will be used to link to sibling page(s) later and, if this is the
+ * initial page in the level, saved in the BTPageState
+ */
+ *blockno = wstate->btws_pages_alloced++;
+
+ /* now allocate and set up the new page */
+ return palloc(BLCKSZ);
+}
+
+static Page
+_bt_bl_alloc_page_shared(BTWriteState *wstate, BlockNumber *blockno, Buffer *buf)
+{
+ /*
+ * Find a shared buffer for the page. Pass mode RBM_ZERO_AND_LOCK to get an
+ * exclusive lock on the buffer content. No lock on the relation as a whole
+ * is needed (as in LockRelationForExtension()) because the initial index
+ * build is not yet complete.
+ */
+ *buf = ReadBufferExtended(wstate->index, MAIN_FORKNUM, P_NEW,
+ RBM_ZERO_AND_LOCK, NULL);
+
+ /*
+ * bufmgr will assign a block number for the new page.
+ * This will be used to link to sibling page(s) later and, if this is the
+ * initial page in the level, saved in the BTPageState
+ */
+ *blockno = BufferGetBlockNumber(*buf);
+
+ return BufferGetPage(*buf);
+}
+
/*
* allocate and initialize a new BTPageState. the returned structure
* is suitable for immediate use by _bt_buildadd.
@@ -688,13 +812,20 @@ _bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
static BTPageState *
_bt_pagestate(BTWriteState *wstate, uint32 level)
{
+ Buffer buf;
+ BlockNumber blockno;
+
BTPageState *state = (BTPageState *) palloc0(sizeof(BTPageState));
- /* create initial page for level */
- state->btps_page = _bt_blnewpage(level);
+ /*
+ * Allocate and initialize initial page for the level, and if using shared
+ * buffers, extend the relation and allocate a shared buffer for the block.
+ */
+ state->btps_page = _bt_blnewpage(level, wstate->_bt_bl_alloc_page(wstate,
+ &blockno, &buf));
- /* and assign it a page position */
- state->btps_blkno = wstate->btws_pages_alloced++;
+ state->btps_blkno = blockno;
+ state->btps_buf = buf;
state->btps_lowkey = NULL;
/* initialize lastoff so first item goes into P_FIRSTKEY */
@@ -829,6 +960,7 @@ _bt_buildadd(BTWriteState *wstate, BTPageState *state, IndexTuple itup,
{
Page npage;
BlockNumber nblkno;
+ Buffer nbuf;
OffsetNumber last_off;
Size last_truncextra;
Size pgspc;
@@ -843,6 +975,7 @@ _bt_buildadd(BTWriteState *wstate, BTPageState *state, IndexTuple itup,
npage = state->btps_page;
nblkno = state->btps_blkno;
+ nbuf = state->btps_buf;
last_off = state->btps_lastoff;
last_truncextra = state->btps_lastextra;
state->btps_lastextra = truncextra;
@@ -899,15 +1032,14 @@ _bt_buildadd(BTWriteState *wstate, BTPageState *state, IndexTuple itup,
*/
Page opage = npage;
BlockNumber oblkno = nblkno;
+ Buffer obuf = nbuf;
ItemId ii;
ItemId hii;
IndexTuple oitup;
- /* Create new page of same level */
- npage = _bt_blnewpage(state->btps_level);
-
- /* and assign it a page position */
- nblkno = wstate->btws_pages_alloced++;
+ /* Create and initialize a new page of same level */
+ npage = _bt_blnewpage(state->btps_level,
+ wstate->_bt_bl_alloc_page(wstate, &nblkno, &nbuf));
/*
* We copy the last item on the page into the new page, and then
@@ -1017,9 +1149,12 @@ _bt_buildadd(BTWriteState *wstate, BTPageState *state, IndexTuple itup,
/*
* Write out the old page. We never need to touch it again, so we can
- * free the opage workspace too.
+ * free the opage workspace too. obuf has been released and is no longer
+ * valid.
*/
- _bt_blwritepage(wstate, opage, oblkno);
+ wstate->_bt_blwritepage(wstate, opage, oblkno, obuf);
+ obuf = InvalidBuffer;
+ opage = NULL;
/*
* Reset last_off to point to new page
@@ -1054,6 +1189,7 @@ _bt_buildadd(BTWriteState *wstate, BTPageState *state, IndexTuple itup,
state->btps_page = npage;
state->btps_blkno = nblkno;
+ state->btps_buf = nbuf;
state->btps_lastoff = last_off;
}
@@ -1099,12 +1235,12 @@ _bt_sort_dedup_finish_pending(BTWriteState *wstate, BTPageState *state,
* Finish writing out the completed btree.
*/
static void
-_bt_uppershutdown(BTWriteState *wstate, BTPageState *state)
+_bt_uppershutdown(BTWriteState *wstate, BTPageState *state, Buffer metabuf,
+ Page metapage)
{
BTPageState *s;
BlockNumber rootblkno = P_NONE;
uint32 rootlevel = 0;
- Page metapage;
/*
* Each iteration of this loop completes one more level of the tree.
@@ -1150,20 +1286,24 @@ _bt_uppershutdown(BTWriteState *wstate, BTPageState *state)
* back one slot. Then we can dump out the page.
*/
_bt_slideleft(s->btps_page);
- _bt_blwritepage(wstate, s->btps_page, s->btps_blkno);
+ wstate->_bt_blwritepage(wstate, s->btps_page, s->btps_blkno, s->btps_buf);
+ s->btps_buf = InvalidBuffer;
s->btps_page = NULL; /* writepage freed the workspace */
}
/*
- * As the last step in the process, construct the metapage and make it
+ * As the last step in the process, initialize the metapage and make it
* point to the new root (unless we had no data at all, in which case it's
* set to point to "P_NONE"). This changes the index to the "valid" state
* by filling in a valid magic number in the metapage.
+ * After this, metapage will have been freed or invalid and metabuf, if ever
+ * valid, will have been released.
*/
- metapage = (Page) palloc(BLCKSZ);
_bt_initmetapage(metapage, rootblkno, rootlevel,
wstate->inskey->allequalimage);
- _bt_blwritepage(wstate, metapage, BTREE_METAPAGE);
+ wstate->_bt_blwritepage(wstate, metapage, BTREE_METAPAGE, metabuf);
+ metabuf = InvalidBuffer;
+ metapage = NULL;
}
/*
@@ -1173,6 +1313,10 @@ _bt_uppershutdown(BTWriteState *wstate, BTPageState *state)
static void
_bt_load(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
{
+ Page metapage;
+ BlockNumber metablkno;
+ Buffer metabuf;
+
BTPageState *state = NULL;
bool merge = (btspool2 != NULL);
IndexTuple itup,
@@ -1185,21 +1329,29 @@ _bt_load(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
int64 tuples_done = 0;
bool deduplicate;
+ deduplicate = wstate->inskey->allequalimage && !btspool->isunique &&
+ BTGetDeduplicateItems(wstate->index);
/*
- * Only bother fsync'ing the data to permanent storage if WAL logging
+ * Reserve block 0 for the metapage up front.
+ *
+ * When using the shared buffers API it is easier to allocate the buffer
+ * for block 0 first instead of trying skip block 0 and allocate it at the
+ * end of index build.
+ *
+ * When not using the shared buffers API, there is no harm in allocating
+ * the metapage first. When block 1 is written, the direct writepage
+ * function will zero-fill block 0. When writing out the metapage at the
+ * end of index build, it will overwrite that block 0.
+ *
+ * The metapage will be initialized and written out at the end of the index
+ * build when all of the information needed to do so is available.
*
- * The self-fsync optimization requires that the backend both add an fsync
- * request to the checkpointer's pending-ops table as well as be prepared
- * to fsync the page data itself. Because none of these are required if the
- * relation is not WAL-logged, pass btws_use_wal for all parameters of the
- * prep function.
+ * The block number will always be BTREE_METAPAGE, so the metablkno
+ * variable is unused and only created to avoid a special case in the
+ * direct alloc function.
*/
- unbuffered_prep(&wstate->ub_wstate, wstate->btws_use_wal,
- wstate->btws_use_wal, wstate->btws_use_wal);
-
- deduplicate = wstate->inskey->allequalimage && !btspool->isunique &&
- BTGetDeduplicateItems(wstate->index);
+ metapage = wstate->_bt_bl_alloc_page(wstate, &metablkno, &metabuf);
if (merge)
{
@@ -1422,10 +1574,7 @@ _bt_load(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
}
/* Close down final pages and write the metapage */
- _bt_uppershutdown(wstate, state);
-
- unbuffered_finish(&wstate->ub_wstate, RelationGetSmgr(wstate->index),
- MAIN_FORKNUM);
+ _bt_uppershutdown(wstate, state, metabuf, metapage);
}
/*
--
2.30.2
[text/x-patch] v5-0003-BTree-index-use-unbuffered-IO-optimization.patch (2.1K, ../../CAAKRu_ZmkCNz-=5U5wZyyi3=Usfs1btsTKc_L9r1ceFvbxE3kg@mail.gmail.com/5-v5-0003-BTree-index-use-unbuffered-IO-optimization.patch)
download | inline diff:
From ed9200e6262e451936cfceb1b41ff079531a83f8 Mon Sep 17 00:00:00 2001
From: Melanie Plageman <[email protected]>
Date: Tue, 8 Feb 2022 19:01:32 -0500
Subject: [PATCH v5 3/4] BTree index use unbuffered IO optimization
While building a btree index, the backend can avoid fsync'ing all of the
pages if it uses the optimization introduced in a prior commit.
This can substantially improve performance when many indexes are being
built during DDL operations.
---
src/backend/access/nbtree/nbtree.c | 2 +-
src/backend/access/nbtree/nbtsort.c | 9 ++++++++-
2 files changed, 9 insertions(+), 2 deletions(-)
diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c
index 843c9e2362..a1efbe1e6a 100644
--- a/src/backend/access/nbtree/nbtree.c
+++ b/src/backend/access/nbtree/nbtree.c
@@ -155,7 +155,7 @@ btbuildempty(Relation index)
Page metapage;
UnBufferedWriteState wstate;
- unbuffered_prep(&wstate, false, true, false);
+ unbuffered_prep(&wstate, true, true, true);
/* Construct metapage. */
metapage = (Page) palloc(BLCKSZ);
diff --git a/src/backend/access/nbtree/nbtsort.c b/src/backend/access/nbtree/nbtsort.c
index a67770f3fd..079832bb78 100644
--- a/src/backend/access/nbtree/nbtsort.c
+++ b/src/backend/access/nbtree/nbtsort.c
@@ -1188,8 +1188,15 @@ _bt_load(BTWriteState *wstate, BTSpool *btspool, BTSpool *btspool2)
/*
* Only bother fsync'ing the data to permanent storage if WAL logging
+ *
+ * The self-fsync optimization requires that the backend both add an fsync
+ * request to the checkpointer's pending-ops table as well as be prepared
+ * to fsync the page data itself. Because none of these are required if the
+ * relation is not WAL-logged, pass btws_use_wal for all parameters of the
+ * prep function.
*/
- unbuffered_prep(&wstate->ub_wstate, false, wstate->btws_use_wal, false);
+ unbuffered_prep(&wstate->ub_wstate, wstate->btws_use_wal,
+ wstate->btws_use_wal, wstate->btws_use_wal);
deduplicate = wstate->inskey->allequalimage && !btspool->isunique &&
BTGetDeduplicateItems(wstate->index);
--
2.30.2
^ permalink raw reply [nested|flat] 152+ messages in thread
* Re: Avoiding smgrimmedsync() during nbtree index builds
2022-01-10 22:50 Re: Avoiding smgrimmedsync() during nbtree index builds Melanie Plageman <[email protected]>
2022-01-11 17:10 ` Re: Avoiding smgrimmedsync() during nbtree index builds Melanie Plageman <[email protected]>
2022-01-23 21:55 ` Re: Avoiding smgrimmedsync() during nbtree index builds Andres Freund <[email protected]>
2022-02-09 18:49 ` Re: Avoiding smgrimmedsync() during nbtree index builds Melanie Plageman <[email protected]>
@ 2022-02-13 14:33 ` Dmitry Dolgov <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Dmitry Dolgov @ 2022-02-13 14:33 UTC (permalink / raw)
To: Melanie Plageman <[email protected]>; +Cc: Andres Freund <[email protected]>; Heikki Linnakangas <[email protected]>; pgsql-hackers; Peter Geoghegan <[email protected]>; Thomas Munro <[email protected]>; Justin Pryzby <[email protected]>
> On Wed, Feb 09, 2022 at 01:49:30PM -0500, Melanie Plageman wrote:
> Hi,
> v5 attached and all email feedback addressed below
Thanks for the patch, it looks quite good.
I don't see it in the discussion, so naturally curious -- why directmgr
is not used for bloom index, e.g. in blbuildempty?
> On Sun, Jan 16, 2022 at 3:26 PM Justin Pryzby <[email protected]> wrote:
> > Separate from this issue, I wonder if it'd be useful to write a DEBUG log
> > showing when btree uses shared_buffers vs fsync. And a regression test which
> > first SETs client_min_messages=debug to capture the debug log to demonstrate
> > when/that new code path is being hit. I'm not sure if that would be good to
> > merge, but it may be useful for now.
I can't find the thread right away, but I vaguely remember a similar
situation where such approach, as a main way to test the patch, had
caused some disagreement. Of course for the development phase it would
be indeed convenient.
^ permalink raw reply [nested|flat] 152+ messages in thread
* Re: gcc 15 "array subscript 0" warning at level -O3
@ 2025-06-04 19:00 Andres Freund <[email protected]>
2025-06-05 19:50 ` Re: gcc 15 "array subscript 0" warning at level -O3 Tom Lane <[email protected]>
2025-07-02 14:13 ` Re: gcc 15 "array subscript 0" warning at level -O3 jian he <[email protected]>
0 siblings, 2 replies; 152+ messages in thread
From: Andres Freund @ 2025-06-04 19:00 UTC (permalink / raw)
To: Tom Lane <[email protected]>; +Cc: [email protected]
Hi,
On 2025-04-25 15:58:29 -0400, Andres Freund wrote:
> On 2025-04-25 13:37:15 -0400, Tom Lane wrote:
> > Buildfarm member serinus has been producing the identical warning for
> > some time. I'd been ignoring that because it runs "experimental gcc",
> > but I guess the experiment has leaked out to production distros.
> >
> > What seems to be happening here is that after inlining
> > assign_simple_var into exec_set_found, the compiler decides that
> > "newvalue" might be zero (since it's a BoolGetDatum result),
> > and then it warns -- in a rather strange way -- about the
> > potential null dereference.
>
> I don't think it actually is complaining about a null dereference - it thinks
> we're interpreting a boolean as a pointer (for which it obviously is not wide
> enough)
>
>
> > The dereference is not reachable
> > because of the preceding "var->datatype->typlen == -1" check,
> > but that's not stopping the optimizer from bitching.
>
> > I experimented with modifying exec_set_found thus:
> >
> > var = (PLpgSQL_var *) (estate->datums[estate->found_varno]);
> > + Assert(var->datatype->typlen == 1);
> > assign_simple_var(estate, var, BoolGetDatum(state), false, false);
> >
> > which should be OK since we're expecting the "found" variable to
> > be boolean. That does silence the warning, but of course only
> > in --enable-cassert builds.
>
> One way to address this is outlined here:
>
> https://postgr.es/m/20230316172818.x6375uvheom3ibt2%40awork3.anarazel.de
> https://postgr.es/m/20240207203138.sknifhlppdtgtxnk%40awork3.anarazel.de
>
> I've been wondering about adding wrapping something like that in a
> pg_assume(expr) or such.
I've been once more annoyed by this warning. Here's a prototype for the
approach outlined above.
Greetings,
Andres Freund
^ permalink raw reply [nested|flat] 152+ messages in thread
* Re: gcc 15 "array subscript 0" warning at level -O3
2025-06-04 19:00 Re: gcc 15 "array subscript 0" warning at level -O3 Andres Freund <[email protected]>
@ 2025-06-05 19:50 ` Tom Lane <[email protected]>
2025-07-09 23:20 ` Re: gcc 15 "array subscript 0" warning at level -O3 Andres Freund <[email protected]>
1 sibling, 1 reply; 152+ messages in thread
From: Tom Lane @ 2025-06-05 19:50 UTC (permalink / raw)
To: Andres Freund <[email protected]>; +Cc: [email protected]
Andres Freund <[email protected]> writes:
>> I've been wondering about adding wrapping something like that in a
>> pg_assume(expr) or such.
> I've been once more annoyed by this warning. Here's a prototype for the
> approach outlined above.
Looks plausible by eyeball. I did notice a typo in the comment:
+ * pg_assume(expr) stats that we assume `expr` to evaluate to true. In assert
s/stats/states/, I think you meant.
regards, tom lane
^ permalink raw reply [nested|flat] 152+ messages in thread
* Re: gcc 15 "array subscript 0" warning at level -O3
2025-06-04 19:00 Re: gcc 15 "array subscript 0" warning at level -O3 Andres Freund <[email protected]>
2025-06-05 19:50 ` Re: gcc 15 "array subscript 0" warning at level -O3 Tom Lane <[email protected]>
@ 2025-07-09 23:20 ` Andres Freund <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Andres Freund @ 2025-07-09 23:20 UTC (permalink / raw)
To: Tom Lane <[email protected]>; +Cc: [email protected]
Hi,
On 2025-06-05 15:50:48 -0400, Tom Lane wrote:
> Andres Freund <[email protected]> writes:
> >> I've been wondering about adding wrapping something like that in a
> >> pg_assume(expr) or such.
>
> > I've been once more annoyed by this warning. Here's a prototype for the
> > approach outlined above.
>
> Looks plausible by eyeball. I did notice a typo in the comment:
>
> + * pg_assume(expr) stats that we assume `expr` to evaluate to true. In assert
>
> s/stats/states/, I think you meant.
Thanks. I pushed it with that tweak and a bit of minor comment burnishing.
Glad to see the last of that warning.
Greetings,
Andres Freund
^ permalink raw reply [nested|flat] 152+ messages in thread
* Re: gcc 15 "array subscript 0" warning at level -O3
2025-06-04 19:00 Re: gcc 15 "array subscript 0" warning at level -O3 Andres Freund <[email protected]>
@ 2025-07-02 14:13 ` jian he <[email protected]>
2025-07-09 23:21 ` Re: gcc 15 "array subscript 0" warning at level -O3 Andres Freund <[email protected]>
1 sibling, 1 reply; 152+ messages in thread
From: jian he @ 2025-07-02 14:13 UTC (permalink / raw)
To: Andres Freund <[email protected]>; +Cc: Tom Lane <[email protected]>; [email protected]
On Thu, Jun 5, 2025 at 3:00 AM Andres Freund <[email protected]> wrote:
>
> > > The dereference is not reachable
> > > because of the preceding "var->datatype->typlen == -1" check,
> > > but that's not stopping the optimizer from bitching.
> >
> > > I experimented with modifying exec_set_found thus:
> > >
> > > var = (PLpgSQL_var *) (estate->datums[estate->found_varno]);
> > > + Assert(var->datatype->typlen == 1);
> > > assign_simple_var(estate, var, BoolGetDatum(state), false, false);
> > >
> > > which should be OK since we're expecting the "found" variable to
> > > be boolean. That does silence the warning, but of course only
> > > in --enable-cassert builds.
> >
> > One way to address this is outlined here:
> >
> > https://postgr.es/m/20230316172818.x6375uvheom3ibt2%40awork3.anarazel.de
> > https://postgr.es/m/20240207203138.sknifhlppdtgtxnk%40awork3.anarazel.de
> >
> > I've been wondering about adding wrapping something like that in a
> > pg_assume(expr) or such.
>
> I've been once more annoyed by this warning. Here's a prototype for the
> approach outlined above.
>
I can confirm the warning disappears when using gcc-14.0 compile
source code with the attached patch.
I didn't review it though.
I didn’t find this in the CommitFest, so I added an entry [1] to make sure it
doesn’t get forgotten...
[1]: https://commitfest.postgresql.org/patch/5888/
^ permalink raw reply [nested|flat] 152+ messages in thread
* Re: gcc 15 "array subscript 0" warning at level -O3
2025-06-04 19:00 Re: gcc 15 "array subscript 0" warning at level -O3 Andres Freund <[email protected]>
2025-07-02 14:13 ` Re: gcc 15 "array subscript 0" warning at level -O3 jian he <[email protected]>
@ 2025-07-09 23:21 ` Andres Freund <[email protected]>
0 siblings, 0 replies; 152+ messages in thread
From: Andres Freund @ 2025-07-09 23:21 UTC (permalink / raw)
To: jian he <[email protected]>; +Cc: Tom Lane <[email protected]>; [email protected]
Hi,
On 2025-07-02 22:13:17 +0800, jian he wrote:
> On Thu, Jun 5, 2025 at 3:00 AM Andres Freund <[email protected]> wrote:
> > I've been once more annoyed by this warning. Here's a prototype for the
> > approach outlined above.
> >
>
> I can confirm the warning disappears when using gcc-14.0 compile
> source code with the attached patch.
> I didn't review it though.
> I didn’t find this in the CommitFest, so I added an entry [1] to make sure it
> doesn’t get forgotten...
Thanks. Pushed it now and closed the CF entry.
Greetings,
Andres Freund
^ permalink raw reply [nested|flat] 152+ messages in thread
end of thread, other threads:[~2025-07-09 23:21 UTC | newest]
Thread overview: 152+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 1/5] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v11] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v11] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v3 1/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v3 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v1 1/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v10 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/2] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v6 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v7 2/7] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v2 2/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v4 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v8 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v5 3/3] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2020-06-07 21:58 [PATCH v9 2/8] Implement CLUSTER of partitioned table.. Justin Pryzby <[email protected]>
2022-01-10 22:50 Re: Avoiding smgrimmedsync() during nbtree index builds Melanie Plageman <[email protected]>
2022-01-11 17:10 ` Re: Avoiding smgrimmedsync() during nbtree index builds Melanie Plageman <[email protected]>
2022-01-13 15:52 ` Re: Avoiding smgrimmedsync() during nbtree index builds Justin Pryzby <[email protected]>
2022-01-23 21:55 ` Re: Avoiding smgrimmedsync() during nbtree index builds Andres Freund <[email protected]>
2022-02-09 18:49 ` Re: Avoiding smgrimmedsync() during nbtree index builds Melanie Plageman <[email protected]>
2022-02-13 14:33 ` Re: Avoiding smgrimmedsync() during nbtree index builds Dmitry Dolgov <[email protected]>
2025-06-04 19:00 Re: gcc 15 "array subscript 0" warning at level -O3 Andres Freund <[email protected]>
2025-06-05 19:50 ` Re: gcc 15 "array subscript 0" warning at level -O3 Tom Lane <[email protected]>
2025-07-09 23:20 ` Re: gcc 15 "array subscript 0" warning at level -O3 Andres Freund <[email protected]>
2025-07-02 14:13 ` Re: gcc 15 "array subscript 0" warning at level -O3 jian he <[email protected]>
2025-07-09 23:21 ` Re: gcc 15 "array subscript 0" warning at level -O3 Andres Freund <[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