agora inbox for [email protected]help / color / mirror / Atom feed
[PATCH 2/2] Add monitoring aid for max_replication_slots. 13+ messages / 2 participants [nested] [flat]
* [PATCH 2/2] Add monitoring aid for max_replication_slots. @ 2017-09-07 10:13 Kyotaro Horiguchi <[email protected]> 0 siblings, 0 replies; 13+ messages in thread From: Kyotaro Horiguchi @ 2017-09-07 10:13 UTC (permalink / raw) Adds two columns "live" and "distance" in pg_replication_slot. Setting max_slot_wal_keep_size, long-disconnected slots may lose sync. The two columns shows how long a slot can live on or how many bytes a slot have lost if max_slot_wal_keep_size is set. --- src/backend/access/transam/xlog.c | 128 ++++++++++++++++++++++++++++++++++- src/backend/catalog/system_views.sql | 4 +- src/backend/replication/slotfuncs.c | 25 ++++++- src/include/access/xlog.h | 1 + src/include/catalog/pg_proc.h | 2 +- src/test/regress/expected/rules.out | 6 +- 6 files changed, 160 insertions(+), 6 deletions(-) diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index cfdae39..be53e0f 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9402,6 +9402,128 @@ CreateRestartPoint(int flags) return true; } + +/* + * Returns the segment number of the oldest file in XLOG directory. + */ +static XLogSegNo +GetOldestXLogFileSegNo(void) +{ + DIR *xldir; + struct dirent *xlde; + XLogSegNo segno = 0; + + xldir = AllocateDir(XLOGDIR); + if (xldir == NULL) + ereport(ERROR, + (errcode_for_file_access(), + errmsg("could not open write-ahead log directory \"%s\": %m", + XLOGDIR))); + + while ((xlde = ReadDir(xldir, XLOGDIR)) != NULL) + { + TimeLineID tli; + XLogSegNo fsegno; + + /* Ignore files that are not XLOG segments */ + if (!IsXLogFileName(xlde->d_name) && + !IsPartialXLogFileName(xlde->d_name)) + continue; + + XLogFromFileName(xlde->d_name, &tli, &fsegno, wal_segment_size); + + /* get minimum segment ignorig timeline ID */ + if (segno == 0 || fsegno < segno) + segno = fsegno; + } + + return segno; +} + +/* + * Check if the record on the given restartLSN is present in XLOG files. + * + * Returns true if it is present. If minSecureLSN is given, it receives the + * LSN at the beginning of the oldest existing WAL segment. + */ +bool +IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minSecureLSN) +{ + XLogRecPtr currpos; + XLogSegNo currSeg; + XLogSegNo restartSeg; + XLogSegNo tailSeg; + XLogSegNo oldestSeg; + uint64 keepSegs; + + currpos = GetXLogWriteRecPtr(); + + SpinLockAcquire(&XLogCtl->info_lck); + oldestSeg = XLogCtl->lastRemovedSegNo; + SpinLockRelease(&XLogCtl->info_lck); + + /* + * oldestSeg is zero before at least one segment has been removed since + * startup. Use oldest segno taken from file names. + */ + if (oldestSeg == 0) + { + static XLogSegNo oldestFileSeg = 0; + + if (oldestFileSeg == 0) + oldestFileSeg = GetOldestXLogFileSegNo(); + /* let it have the same meaning with lastRemovedSegNo here */ + oldestSeg = oldestFileSeg - 1; + } + + /* oldest segment is just after the last removed segment */ + oldestSeg++; + + XLByteToSeg(currpos, currSeg, wal_segment_size); + XLByteToSeg(restartLSN, restartSeg, wal_segment_size); + + + if (minSecureLSN) + { + if (max_slot_wal_keep_size_mb > 0) + { + uint64 slotlimitbytes = 1024 * 1024 * max_slot_wal_keep_size_mb; + uint64 slotlimitfragment = XLogSegmentOffset(slotlimitbytes, + wal_segment_size); + uint64 currposoff = XLogSegmentOffset(currpos, wal_segment_size); + + /* Calculate keep segments. Must be in sync with KeepLogSeg. */ + Assert(wal_keep_segments >= 0); + Assert(max_slot_wal_keep_size_mb >= 0); + + keepSegs = wal_keep_segments + + ConvertToXSegs(max_slot_wal_keep_size_mb, wal_segment_size); + if (currposoff < slotlimitfragment) + keepSegs++; + + /* + * calculate the oldest segment that will be kept by + * wal_keep_segments and max_slot_wal_keep_size_mb + */ + if (currSeg < keepSegs) + tailSeg = 0; + else + tailSeg = currSeg - keepSegs; + + } + else + { + /* all requred segments are secured in this case */ + XLogRecPtr keep = XLogGetReplicationSlotMinimumLSN(); + XLByteToSeg(keep, tailSeg, wal_segment_size); + } + + XLogSegNoOffsetToRecPtr(tailSeg, 0, *minSecureLSN, wal_segment_size); + } + + return oldestSeg <= restartSeg; +} + /* * Retreat *logSegNo to the last segment that we need to retain because of * either wal_keep_segments or replication slots. @@ -9429,7 +9551,11 @@ KeepLogSeg(XLogRecPtr recptr, XLogSegNo *logSegNo) segno = segno - wal_keep_segments; } - /* then check whether slots limit removal further */ + /* + * then check whether slots limit removal further + * should be consistent with IsLsnStillAvaiable(). + */ + if (max_replication_slots > 0 && keep != InvalidXLogRecPtr) { XLogSegNo slotSegNo; diff --git a/src/backend/catalog/system_views.sql b/src/backend/catalog/system_views.sql index dc40cde..6512ac3 100644 --- a/src/backend/catalog/system_views.sql +++ b/src/backend/catalog/system_views.sql @@ -793,7 +793,9 @@ CREATE VIEW pg_replication_slots AS L.xmin, L.catalog_xmin, L.restart_lsn, - L.confirmed_flush_lsn + L.confirmed_flush_lsn, + L.status, + L.min_secure_lsn FROM pg_get_replication_slots() AS L LEFT JOIN pg_database D ON (L.datoid = D.oid); diff --git a/src/backend/replication/slotfuncs.c b/src/backend/replication/slotfuncs.c index ab776e8..200a478 100644 --- a/src/backend/replication/slotfuncs.c +++ b/src/backend/replication/slotfuncs.c @@ -182,7 +182,7 @@ pg_drop_replication_slot(PG_FUNCTION_ARGS) Datum pg_get_replication_slots(PG_FUNCTION_ARGS) { -#define PG_GET_REPLICATION_SLOTS_COLS 11 +#define PG_GET_REPLICATION_SLOTS_COLS 13 ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; TupleDesc tupdesc; Tuplestorestate *tupstore; @@ -304,6 +304,29 @@ pg_get_replication_slots(PG_FUNCTION_ARGS) else nulls[i++] = true; + if (restart_lsn == InvalidXLogRecPtr) + { + values[i++] = CStringGetTextDatum("unknown"); + values[i++] = LSNGetDatum(InvalidXLogRecPtr); + } + else + { + XLogRecPtr min_secure_lsn; + char *status = "borken"; + + if (BoolGetDatum(IsLsnStillAvaiable(restart_lsn, + &min_secure_lsn))) + { + if (min_secure_lsn <= restart_lsn) + status = "secured"; + else + status = "insecured"; + } + + values[i++] = CStringGetTextDatum(status); + values[i++] = LSNGetDatum(min_secure_lsn); + } + tuplestore_putvalues(tupstore, tupdesc, values, nulls); } LWLockRelease(ReplicationSlotControlLock); diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h index f0c0255..a316ead 100644 --- a/src/include/access/xlog.h +++ b/src/include/access/xlog.h @@ -269,6 +269,7 @@ extern void ShutdownXLOG(int code, Datum arg); extern void InitXLOGAccess(void); extern void CreateCheckPoint(int flags); extern bool CreateRestartPoint(int flags); +extern bool IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minSecureLSN); extern void XLogPutNextOid(Oid nextOid); extern XLogRecPtr XLogRestorePoint(const char *rpName); extern void UpdateFullPageWrites(void); diff --git a/src/include/catalog/pg_proc.h b/src/include/catalog/pg_proc.h index 93c031a..d03fd6f 100644 --- a/src/include/catalog/pg_proc.h +++ b/src/include/catalog/pg_proc.h @@ -5340,7 +5340,7 @@ DATA(insert OID = 3779 ( pg_create_physical_replication_slot PGNSP PGUID 12 1 0 DESCR("create a physical replication slot"); DATA(insert OID = 3780 ( pg_drop_replication_slot PGNSP PGUID 12 1 0 0 0 f f f f t f v u 1 0 2278 "19" _null_ _null_ _null_ _null_ _null_ pg_drop_replication_slot _null_ _null_ _null_ )); DESCR("drop a replication slot"); -DATA(insert OID = 3781 ( pg_get_replication_slots PGNSP PGUID 12 1 10 0 0 f f f f f t s s 0 0 2249 "" "{19,19,25,26,16,16,23,28,28,3220,3220}" "{o,o,o,o,o,o,o,o,o,o,o}" "{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn}" _null_ _null_ pg_get_replication_slots _null_ _null_ _null_ )); +DATA(insert OID = 3781 ( pg_get_replication_slots PGNSP PGUID 12 1 10 0 0 f f f f f t s s 0 0 2249 "" "{19,19,25,26,16,16,23,28,28,3220,3220,25,3220}" "{o,o,o,o,o,o,o,o,o,o,o,o,o}" "{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn,status,min_secure_lsn}" _null_ _null_ pg_get_replication_slots _null_ _null_ _null_ )); DESCR("information about replication slots currently in use"); DATA(insert OID = 3786 ( pg_create_logical_replication_slot PGNSP PGUID 12 1 0 0 0 f f f f t f v u 3 0 2249 "19 19 16" "{19,19,16,25,3220}" "{i,i,i,o,o}" "{slot_name,plugin,temporary,slot_name,lsn}" _null_ _null_ pg_create_logical_replication_slot _null_ _null_ _null_ )); DESCR("set up a logical replication slot"); diff --git a/src/test/regress/expected/rules.out b/src/test/regress/expected/rules.out index f1c1b44..d9d74a3 100644 --- a/src/test/regress/expected/rules.out +++ b/src/test/regress/expected/rules.out @@ -1451,8 +1451,10 @@ pg_replication_slots| SELECT l.slot_name, l.xmin, l.catalog_xmin, l.restart_lsn, - l.confirmed_flush_lsn - FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn) + l.confirmed_flush_lsn, + l.status, + l.min_secure_lsn + FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn, status, min_secure_lsn) LEFT JOIN pg_database d ON ((l.datoid = d.oid))); pg_roles| SELECT pg_authid.rolname, pg_authid.rolsuper, -- 2.9.2 ----Next_Part(Thu_Nov_09_17_31_28_2017_905)-- Content-Type: text/plain Content-Disposition: inline Content-Transfer-Encoding: 8bit MIME-Version: 1.0 -- Sent via pgsql-hackers mailing list ([email protected]) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers ----Next_Part(Thu_Nov_09_17_31_28_2017_905)---- ^ permalink raw reply [nested|flat] 13+ messages in thread
* [PATCH 2/2] Add monitoring aid for max_replication_slots. @ 2017-09-07 10:13 Kyotaro Horiguchi <[email protected]> 0 siblings, 0 replies; 13+ messages in thread From: Kyotaro Horiguchi @ 2017-09-07 10:13 UTC (permalink / raw) Adds two columns "live" and "distance" in pg_replication_slot. Setting max_slot_wal_keep_size, long-disconnected slots may lose sync. The two columns shows how long a slot can live on or how many bytes a slot have lost if max_slot_wal_keep_size is set. --- src/backend/access/transam/xlog.c | 122 ++++++++++++++++++++++++++++++++++- src/backend/catalog/system_views.sql | 4 +- src/backend/replication/slotfuncs.c | 27 +++++++- src/include/access/xlog.h | 1 + src/include/catalog/pg_proc.h | 2 +- src/test/regress/expected/rules.out | 6 +- 6 files changed, 156 insertions(+), 6 deletions(-) diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index cfdae39..8ce7044 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9402,6 +9402,122 @@ CreateRestartPoint(int flags) return true; } + +/* + * Returns the segment number of the oldest file in XLOG directory. + */ +static XLogSegNo +GetOldestXLogFileSegNo(void) +{ + DIR *xldir; + struct dirent *xlde; + XLogSegNo segno = 0; + + xldir = AllocateDir(XLOGDIR); + if (xldir == NULL) + ereport(ERROR, + (errcode_for_file_access(), + errmsg("could not open write-ahead log directory \"%s\": %m", + XLOGDIR))); + + while ((xlde = ReadDir(xldir, XLOGDIR)) != NULL) + { + TimeLineID tli; + XLogSegNo fsegno; + + /* Ignore files that are not XLOG segments */ + if (!IsXLogFileName(xlde->d_name) && + !IsPartialXLogFileName(xlde->d_name)) + continue; + + XLogFromFileName(xlde->d_name, &tli, &fsegno, wal_segment_size); + + /* get minimum segment ignorig timeline ID */ + if (segno == 0 || fsegno < segno) + segno = fsegno; + } + + return segno; +} + +/* + * Check if the record on the given restartLSN is present in XLOG files. + * + * Returns true if it is present. If distance is given, it receives the + * distance to the point where the margin defined by max_slot_wal_keep_size_mb + * and wal_keep_segments will be exhausted or how many bytes we have lost + * after restartLSN. + * + * true and distance = 0 means that restartLSN will be lost by at most two + * checkpoints. + */ +bool +IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minSecureLSN) +{ + XLogRecPtr currpos; + XLogSegNo currSeg; + XLogSegNo restartSeg; + XLogSegNo tailSeg; + XLogSegNo oldestSeg; + uint64 keepSegs; + + currpos = GetXLogWriteRecPtr(); + + SpinLockAcquire(&XLogCtl->info_lck); + oldestSeg = XLogCtl->lastRemovedSegNo; + SpinLockRelease(&XLogCtl->info_lck); + + /* + * oldestSeg is zero before at least one segment has been removed since + * startup. Use oldest segno taken from file names. + */ + if (oldestSeg == 0) + { + static XLogSegNo oldestFileSeg = 0; + + if (oldestFileSeg == 0) + oldestFileSeg = GetOldestXLogFileSegNo(); + /* let it have the same meaning with lastRemovedSegNo here */ + oldestSeg = oldestFileSeg - 1; + } + + /* oldest segment is just after the last removed segment */ + oldestSeg++; + + XLByteToSeg(currpos, currSeg, wal_segment_size); + XLByteToSeg(restartLSN, restartSeg, wal_segment_size); + + + if (minSecureLSN) + { + uint64 slotlimitbytes = 1024 * 1024 * max_slot_wal_keep_size_mb; + uint64 slotlimitfragment = XLogSegmentOffset(slotlimitbytes, + wal_segment_size); + uint64 currposoff = XLogSegmentOffset(currpos, wal_segment_size); + + /* Calculate keep segments. Must be in sync with KeepLogSeg. */ + Assert(wal_keep_segments >= 0); + Assert(max_slot_wal_keep_size_mb >= 0); + keepSegs = wal_keep_segments + + ConvertToXSegs(max_slot_wal_keep_size_mb, wal_segment_size); + if (currposoff < slotlimitfragment) + keepSegs++; + + /* + * calculate the oldest segment that will be kept by wal_keep_segments and + * max_slot_wal_keep_size_mb + */ + if (currSeg < keepSegs) + tailSeg = 0; + else + tailSeg = currSeg - keepSegs; + + XLogSegNoOffsetToRecPtr(tailSeg, 0, *minSecureLSN, wal_segment_size); + } + + return oldestSeg <= restartSeg; +} + /* * Retreat *logSegNo to the last segment that we need to retain because of * either wal_keep_segments or replication slots. @@ -9429,7 +9545,11 @@ KeepLogSeg(XLogRecPtr recptr, XLogSegNo *logSegNo) segno = segno - wal_keep_segments; } - /* then check whether slots limit removal further */ + /* + * then check whether slots limit removal further + * should be consistent with IsLsnStillAvaiable(). + */ + if (max_replication_slots > 0 && keep != InvalidXLogRecPtr) { XLogSegNo slotSegNo; diff --git a/src/backend/catalog/system_views.sql b/src/backend/catalog/system_views.sql index dc40cde..6512ac3 100644 --- a/src/backend/catalog/system_views.sql +++ b/src/backend/catalog/system_views.sql @@ -793,7 +793,9 @@ CREATE VIEW pg_replication_slots AS L.xmin, L.catalog_xmin, L.restart_lsn, - L.confirmed_flush_lsn + L.confirmed_flush_lsn, + L.status, + L.min_secure_lsn FROM pg_get_replication_slots() AS L LEFT JOIN pg_database D ON (L.datoid = D.oid); diff --git a/src/backend/replication/slotfuncs.c b/src/backend/replication/slotfuncs.c index ab776e8..0dca618 100644 --- a/src/backend/replication/slotfuncs.c +++ b/src/backend/replication/slotfuncs.c @@ -182,7 +182,7 @@ pg_drop_replication_slot(PG_FUNCTION_ARGS) Datum pg_get_replication_slots(PG_FUNCTION_ARGS) { -#define PG_GET_REPLICATION_SLOTS_COLS 11 +#define PG_GET_REPLICATION_SLOTS_COLS 13 ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; TupleDesc tupdesc; Tuplestorestate *tupstore; @@ -304,6 +304,31 @@ pg_get_replication_slots(PG_FUNCTION_ARGS) else nulls[i++] = true; + if (max_slot_wal_keep_size_mb > 0 && restart_lsn != InvalidXLogRecPtr) + { + XLogRecPtr min_secure_lsn; + char *status = "unknown"; + + if (BoolGetDatum(IsLsnStillAvaiable(restart_lsn, + &min_secure_lsn))) + { + if (min_secure_lsn <= restart_lsn) + status = "secured"; + else + status = "insecured"; + } + else + status = "broken"; + + values[i++] = CStringGetTextDatum(status); + values[i++] = LSNGetDatum(min_secure_lsn); + } + else + { + values[i++] = BoolGetDatum(true); + nulls[i++] = true; + } + tuplestore_putvalues(tupstore, tupdesc, values, nulls); } LWLockRelease(ReplicationSlotControlLock); diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h index f0c0255..a316ead 100644 --- a/src/include/access/xlog.h +++ b/src/include/access/xlog.h @@ -269,6 +269,7 @@ extern void ShutdownXLOG(int code, Datum arg); extern void InitXLOGAccess(void); extern void CreateCheckPoint(int flags); extern bool CreateRestartPoint(int flags); +extern bool IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minSecureLSN); extern void XLogPutNextOid(Oid nextOid); extern XLogRecPtr XLogRestorePoint(const char *rpName); extern void UpdateFullPageWrites(void); diff --git a/src/include/catalog/pg_proc.h b/src/include/catalog/pg_proc.h index 93c031a..d03fd6f 100644 --- a/src/include/catalog/pg_proc.h +++ b/src/include/catalog/pg_proc.h @@ -5340,7 +5340,7 @@ DATA(insert OID = 3779 ( pg_create_physical_replication_slot PGNSP PGUID 12 1 0 DESCR("create a physical replication slot"); DATA(insert OID = 3780 ( pg_drop_replication_slot PGNSP PGUID 12 1 0 0 0 f f f f t f v u 1 0 2278 "19" _null_ _null_ _null_ _null_ _null_ pg_drop_replication_slot _null_ _null_ _null_ )); DESCR("drop a replication slot"); -DATA(insert OID = 3781 ( pg_get_replication_slots PGNSP PGUID 12 1 10 0 0 f f f f f t s s 0 0 2249 "" "{19,19,25,26,16,16,23,28,28,3220,3220}" "{o,o,o,o,o,o,o,o,o,o,o}" "{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn}" _null_ _null_ pg_get_replication_slots _null_ _null_ _null_ )); +DATA(insert OID = 3781 ( pg_get_replication_slots PGNSP PGUID 12 1 10 0 0 f f f f f t s s 0 0 2249 "" "{19,19,25,26,16,16,23,28,28,3220,3220,25,3220}" "{o,o,o,o,o,o,o,o,o,o,o,o,o}" "{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn,status,min_secure_lsn}" _null_ _null_ pg_get_replication_slots _null_ _null_ _null_ )); DESCR("information about replication slots currently in use"); DATA(insert OID = 3786 ( pg_create_logical_replication_slot PGNSP PGUID 12 1 0 0 0 f f f f t f v u 3 0 2249 "19 19 16" "{19,19,16,25,3220}" "{i,i,i,o,o}" "{slot_name,plugin,temporary,slot_name,lsn}" _null_ _null_ pg_create_logical_replication_slot _null_ _null_ _null_ )); DESCR("set up a logical replication slot"); diff --git a/src/test/regress/expected/rules.out b/src/test/regress/expected/rules.out index f1c1b44..d9d74a3 100644 --- a/src/test/regress/expected/rules.out +++ b/src/test/regress/expected/rules.out @@ -1451,8 +1451,10 @@ pg_replication_slots| SELECT l.slot_name, l.xmin, l.catalog_xmin, l.restart_lsn, - l.confirmed_flush_lsn - FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn) + l.confirmed_flush_lsn, + l.status, + l.min_secure_lsn + FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn, status, min_secure_lsn) LEFT JOIN pg_database d ON ((l.datoid = d.oid))); pg_roles| SELECT pg_authid.rolname, pg_authid.rolsuper, -- 2.9.2 ----Next_Part(Wed_Nov_08_13_14_31_2017_781)-- Content-Type: text/plain Content-Disposition: inline Content-Transfer-Encoding: 8bit MIME-Version: 1.0 -- Sent via pgsql-hackers mailing list ([email protected]) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers ----Next_Part(Wed_Nov_08_13_14_31_2017_781)---- ^ permalink raw reply [nested|flat] 13+ messages in thread
* [PATCH 2/2] Add monitoring aid for max_replication_slots. @ 2017-09-07 10:13 Kyotaro Horiguchi <[email protected]> 0 siblings, 0 replies; 13+ messages in thread From: Kyotaro Horiguchi @ 2017-09-07 10:13 UTC (permalink / raw) Adds two columns "live" and "distance" in pg_replication_slot. Setting max_slot_wal_keep_size, long-disconnected slots may lose sync. The two columns shows how long a slot can live on or how many bytes a slot have lost if max_slot_wal_keep_size is set. --- src/backend/access/transam/xlog.c | 80 ++++++++++++++++++++++++++++++++++++ src/backend/catalog/system_views.sql | 4 +- src/backend/replication/slotfuncs.c | 16 +++++++- src/include/access/xlog.h | 1 + src/include/catalog/pg_proc.h | 2 +- 5 files changed, 100 insertions(+), 3 deletions(-) diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index ae70d7d..c4c8307 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9324,6 +9324,86 @@ CreateRestartPoint(int flags) } /* + * Check if the record on the given lsn will be preserved at the next + * checkpoint. + * + * Returns true if it will be preserved. If distance is given, the distance + * from origin to the beginning of the first segment kept at the next + * checkpoint. It means margin when this function returns true and gap of lost + * records when false. + * + * This function should return the consistent result with KeepLogSeg. + */ +bool +GetMarginToSlotSegmentLimit(XLogRecPtr restartLSN, uint64 *distance) +{ + XLogRecPtr currpos; + XLogRecPtr tailpos; + uint64 currSeg; + uint64 restByteInSeg; + uint64 restartSeg; + uint64 tailSeg; + uint64 keepSegs; + + currpos = GetXLogWriteRecPtr(); + + LWLockAcquire(ControlFileLock, LW_SHARED); + tailpos = ControlFile->checkPointCopy.redo; + LWLockRelease(ControlFileLock); + + /* Move the pointer to the beginning of the segment*/ + XLByteToSeg(currpos, currSeg); + XLByteToSeg(restartLSN, restartSeg); + XLByteToSeg(tailpos, tailSeg); + restByteInSeg = 0; + + Assert(wal_keep_segments >= 0); + Assert(max_slot_wal_keep_size_mb >= 0); + + /* + * WAL are removed by the unit of segment. + */ + keepSegs = wal_keep_segments + ConvertToXSegs(max_slot_wal_keep_size_mb); + + /* + * If the latest checkpoint's redo point is older than the current head + * minus keep segments, the next checkpoint keeps the redo point's + * segment. Elsewise use current head minus number of segments to keep. + */ + if (currSeg < tailSeg + keepSegs) + { + if (currSeg < keepSegs) + tailSeg = 0; + else + tailSeg = currSeg - keepSegs; + + /* In this case, the margin will be the bytes to the next segment */ + restByteInSeg = XLogSegSize - (currpos % XLogSegSize); + } + + elog(LOG, "rest %lx, curr %lx, tail %lx", restartSeg, currSeg, tailSeg); + /* Required sements will be removed at the next checkpoint */ + if (restartSeg < tailSeg) + { + /* Calculate how may bytes the slot have lost */ + if (distance) + { + uint64 restbytes = (restartSeg + 1) * XLogSegSize - restartLSN; + *distance = + (tailSeg - restartSeg - 1) * XLogSegSize + + restbytes; + } + return false; + } + + /* Margin at the next checkpoint before the slot lose sync */ + if (distance) + *distance = (restartSeg - tailSeg) * XLogSegSize + restByteInSeg; + + return true; +} + +/* * Retreat *logSegNo to the last segment that we need to retain because of * either wal_keep_segments or replication slots. * diff --git a/src/backend/catalog/system_views.sql b/src/backend/catalog/system_views.sql index dc40cde..c55c88b 100644 --- a/src/backend/catalog/system_views.sql +++ b/src/backend/catalog/system_views.sql @@ -793,7 +793,9 @@ CREATE VIEW pg_replication_slots AS L.xmin, L.catalog_xmin, L.restart_lsn, - L.confirmed_flush_lsn + L.confirmed_flush_lsn, + L.live, + L.distance FROM pg_get_replication_slots() AS L LEFT JOIN pg_database D ON (L.datoid = D.oid); diff --git a/src/backend/replication/slotfuncs.c b/src/backend/replication/slotfuncs.c index d4cbd83..52a9d26 100644 --- a/src/backend/replication/slotfuncs.c +++ b/src/backend/replication/slotfuncs.c @@ -182,7 +182,7 @@ pg_drop_replication_slot(PG_FUNCTION_ARGS) Datum pg_get_replication_slots(PG_FUNCTION_ARGS) { -#define PG_GET_REPLICATION_SLOTS_COLS 11 +#define PG_GET_REPLICATION_SLOTS_COLS 13 ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; TupleDesc tupdesc; Tuplestorestate *tupstore; @@ -304,6 +304,20 @@ pg_get_replication_slots(PG_FUNCTION_ARGS) else nulls[i++] = true; + if (max_slot_wal_keep_size_mb > 0 && restart_lsn != InvalidXLogRecPtr) + { + uint64 distance; + + values[i++] = BoolGetDatum(GetMarginToSlotSegmentLimit(restart_lsn, + &distance)); + values[i++] = Int64GetDatum(distance); + } + else + { + values[i++] = BoolGetDatum(true); + nulls[i++] = true; + } + tuplestore_putvalues(tupstore, tupdesc, values, nulls); } LWLockRelease(ReplicationSlotControlLock); diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h index f596794..30507eb 100644 --- a/src/include/access/xlog.h +++ b/src/include/access/xlog.h @@ -267,6 +267,7 @@ extern void ShutdownXLOG(int code, Datum arg); extern void InitXLOGAccess(void); extern void CreateCheckPoint(int flags); extern bool CreateRestartPoint(int flags); +extern bool GetMarginToSlotSegmentLimit(XLogRecPtr restartLSN, uint64 *distance); extern void XLogPutNextOid(Oid nextOid); extern XLogRecPtr XLogRestorePoint(const char *rpName); extern void UpdateFullPageWrites(void); diff --git a/src/include/catalog/pg_proc.h b/src/include/catalog/pg_proc.h index 8b33b4e..4bad345 100644 --- a/src/include/catalog/pg_proc.h +++ b/src/include/catalog/pg_proc.h @@ -5293,7 +5293,7 @@ DATA(insert OID = 3779 ( pg_create_physical_replication_slot PGNSP PGUID 12 1 0 DESCR("create a physical replication slot"); DATA(insert OID = 3780 ( pg_drop_replication_slot PGNSP PGUID 12 1 0 0 0 f f f f t f v u 1 0 2278 "19" _null_ _null_ _null_ _null_ _null_ pg_drop_replication_slot _null_ _null_ _null_ )); DESCR("drop a replication slot"); -DATA(insert OID = 3781 ( pg_get_replication_slots PGNSP PGUID 12 1 10 0 0 f f f f f t s s 0 0 2249 "" "{19,19,25,26,16,16,23,28,28,3220,3220}" "{o,o,o,o,o,o,o,o,o,o,o}" "{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn}" _null_ _null_ pg_get_replication_slots _null_ _null_ _null_ )); +DATA(insert OID = 3781 ( pg_get_replication_slots PGNSP PGUID 12 1 10 0 0 f f f f f t s s 0 0 2249 "" "{19,19,25,26,16,16,23,28,28,3220,3220,16,3220}" "{o,o,o,o,o,o,o,o,o,o,o,o,o}" "{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn,live,distance}" _null_ _null_ pg_get_replication_slots _null_ _null_ _null_ )); DESCR("information about replication slots currently in use"); DATA(insert OID = 3786 ( pg_create_logical_replication_slot PGNSP PGUID 12 1 0 0 0 f f f f t f v u 3 0 2249 "19 19 16" "{19,19,16,25,3220}" "{i,i,i,o,o}" "{slot_name,plugin,temporary,slot_name,lsn}" _null_ _null_ pg_create_logical_replication_slot _null_ _null_ _null_ )); DESCR("set up a logical replication slot"); -- 2.9.2 ----Next_Part(Thu_Sep_07_21_59_56_2017_119)-- Content-Type: text/plain Content-Disposition: inline Content-Transfer-Encoding: 8bit MIME-Version: 1.0 -- Sent via pgsql-hackers mailing list ([email protected]) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers ----Next_Part(Thu_Sep_07_21_59_56_2017_119)---- ^ permalink raw reply [nested|flat] 13+ messages in thread
* [PATCH 2/2] Add monitoring aid for max_replication_slots. @ 2017-09-07 10:13 Kyotaro Horiguchi <[email protected]> 0 siblings, 0 replies; 13+ messages in thread From: Kyotaro Horiguchi @ 2017-09-07 10:13 UTC (permalink / raw) Adds two columns "live" and "distance" in pg_replication_slot. Setting max_slot_wal_keep_size, long-disconnected slots may lose sync. The two columns shows how long a slot can live on or how many bytes a slot have lost if max_slot_wal_keep_size is set. --- src/backend/access/transam/xlog.c | 137 ++++++++++++++++++++++++++++++++++- src/backend/catalog/system_views.sql | 4 +- src/backend/replication/slotfuncs.c | 16 +++- src/include/access/xlog.h | 1 + src/include/catalog/pg_proc.h | 2 +- src/test/regress/expected/rules.out | 6 +- 6 files changed, 160 insertions(+), 6 deletions(-) diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index f79cefb..a9203ff 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9402,6 +9402,140 @@ CreateRestartPoint(int flags) return true; } + +/* + * Returns the segment number of the oldest file in XLOG directory. + */ +static XLogSegNo +GetOldestXLogFileSegNo(void) +{ + DIR *xldir; + struct dirent *xlde; + XLogSegNo segno = 0; + + xldir = AllocateDir(XLOGDIR); + if (xldir == NULL) + ereport(ERROR, + (errcode_for_file_access(), + errmsg("could not open write-ahead log directory \"%s\": %m", + XLOGDIR))); + + while ((xlde = ReadDir(xldir, XLOGDIR)) != NULL) + { + TimeLineID tli; + XLogSegNo fsegno; + + /* Ignore files that are not XLOG segments */ + if (!IsXLogFileName(xlde->d_name) && + !IsPartialXLogFileName(xlde->d_name)) + continue; + + XLogFromFileName(xlde->d_name, &tli, &fsegno, wal_segment_size); + + /* get minimum segment ignorig timeline ID */ + if (segno == 0 || fsegno < segno) + segno = fsegno; + } + + return segno; +} + +/* + * Check if the record on the given restartLSN is present in XLOG files. + * + * Returns true if it is present. If distance is given, it receives the + * distance to the point where the margin defined by max_slot_wal_keep_size_mb + * and wal_keep_segments will be exhausted or how many bytes we have lost + * after restartLSN. + * + * true and distance = 0 means that restartLSN will be lost by at most two + * checkpoints. + */ +bool +IsLsnStillAvaiable(XLogRecPtr restartLSN, uint64 *distance) +{ + XLogRecPtr currpos; + XLogSegNo currSeg; + XLogSegNo restartSeg; + XLogSegNo tailSeg; + XLogSegNo oldestSeg; + uint64 keepSegs; + uint64 restbytes; + + + currpos = GetXLogWriteRecPtr(); + + SpinLockAcquire(&XLogCtl->info_lck); + oldestSeg = XLogCtl->lastRemovedSegNo; + SpinLockRelease(&XLogCtl->info_lck); + + /* + * oldestSeg is zero before at least one segment has been removed since + * startup. Use oldest segno taken from file names. + */ + if (oldestSeg == 0) + { + static XLogSegNo oldestFileSeg = 0; + + if (oldestFileSeg == 0) + oldestFileSeg = GetOldestXLogFileSegNo(); + /* let it have the same meaning with lastRemovedSegNo here */ + oldestSeg = oldestFileSeg - 1; + } + + /* oldest segment is just after the last removed segment */ + oldestSeg++; + + XLByteToSeg(currpos, currSeg, wal_segment_size); + XLByteToSeg(restartLSN, restartSeg, wal_segment_size); + + /* no need to calculate distance. very easy. */ + if (!distance) + return oldestSeg <= restartSeg; + + /* This must perform the same thing as KeepLogSeg. */ + Assert(wal_keep_segments >= 0); + Assert(max_slot_wal_keep_size_mb >= 0); + keepSegs = wal_keep_segments + + ConvertToXSegs(max_slot_wal_keep_size_mb, wal_segment_size); + + /* + * calculate the oldest segment that will be kept by wal_keep_segments and + * max_slot_wal_keep_size_mb + */ + if (currSeg < keepSegs) + tailSeg = 0; + else + tailSeg = currSeg - keepSegs; + + if (restartSeg < oldestSeg) + { + /* + * restartSeg has been removed. Calculate how many bytes from the + * restartLSN have lost. + */ + restbytes = (restartSeg + 1) * wal_segment_size - restartLSN; + *distance = + (oldestSeg - (restartSeg + 1)) * wal_segment_size + restbytes; + + return false; + } + + if (tailSeg <= restartSeg) + { + /* Return how many bytes we can advance before the slot loses margin */ + restbytes = wal_segment_size - (currpos % wal_segment_size); + *distance = (restartSeg - tailSeg) * wal_segment_size + restbytes; + } + else + { + /* the margin ran out */ + *distance = 0; + } + + return true; +} + /* * Retreat *logSegNo to the last segment that we need to retain because of * either wal_keep_segments or replication slots. @@ -9433,7 +9567,8 @@ KeepLogSeg(XLogRecPtr recptr, XLogSegNo *logSegNo) if (max_replication_slots > 0 && keep != InvalidXLogRecPtr) { XLogSegNo slotSegNo; - int slotlimitsegs = ConvertToXSegs(max_slot_wal_keep_size_mb); + int slotlimitsegs = + ConvertToXSegs(max_slot_wal_keep_size_mb, wal_segment_size); XLByteToSeg(keep, slotSegNo, wal_segment_size); diff --git a/src/backend/catalog/system_views.sql b/src/backend/catalog/system_views.sql index dc40cde..c55c88b 100644 --- a/src/backend/catalog/system_views.sql +++ b/src/backend/catalog/system_views.sql @@ -793,7 +793,9 @@ CREATE VIEW pg_replication_slots AS L.xmin, L.catalog_xmin, L.restart_lsn, - L.confirmed_flush_lsn + L.confirmed_flush_lsn, + L.live, + L.distance FROM pg_get_replication_slots() AS L LEFT JOIN pg_database D ON (L.datoid = D.oid); diff --git a/src/backend/replication/slotfuncs.c b/src/backend/replication/slotfuncs.c index ab776e8..107da1a 100644 --- a/src/backend/replication/slotfuncs.c +++ b/src/backend/replication/slotfuncs.c @@ -182,7 +182,7 @@ pg_drop_replication_slot(PG_FUNCTION_ARGS) Datum pg_get_replication_slots(PG_FUNCTION_ARGS) { -#define PG_GET_REPLICATION_SLOTS_COLS 11 +#define PG_GET_REPLICATION_SLOTS_COLS 13 ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; TupleDesc tupdesc; Tuplestorestate *tupstore; @@ -304,6 +304,20 @@ pg_get_replication_slots(PG_FUNCTION_ARGS) else nulls[i++] = true; + if (max_slot_wal_keep_size_mb > 0 && restart_lsn != InvalidXLogRecPtr) + { + uint64 distance; + + values[i++] = BoolGetDatum(IsLsnStillAvaiable(restart_lsn, + &distance)); + values[i++] = Int64GetDatum(distance); + } + else + { + values[i++] = BoolGetDatum(true); + nulls[i++] = true; + } + tuplestore_putvalues(tupstore, tupdesc, values, nulls); } LWLockRelease(ReplicationSlotControlLock); diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h index f0c0255..a7a1e4d 100644 --- a/src/include/access/xlog.h +++ b/src/include/access/xlog.h @@ -269,6 +269,7 @@ extern void ShutdownXLOG(int code, Datum arg); extern void InitXLOGAccess(void); extern void CreateCheckPoint(int flags); extern bool CreateRestartPoint(int flags); +extern bool IsLsnStillAvaiable(XLogRecPtr restartLSN, uint64 *distance); extern void XLogPutNextOid(Oid nextOid); extern XLogRecPtr XLogRestorePoint(const char *rpName); extern void UpdateFullPageWrites(void); diff --git a/src/include/catalog/pg_proc.h b/src/include/catalog/pg_proc.h index 93c031a..0913e56 100644 --- a/src/include/catalog/pg_proc.h +++ b/src/include/catalog/pg_proc.h @@ -5340,7 +5340,7 @@ DATA(insert OID = 3779 ( pg_create_physical_replication_slot PGNSP PGUID 12 1 0 DESCR("create a physical replication slot"); DATA(insert OID = 3780 ( pg_drop_replication_slot PGNSP PGUID 12 1 0 0 0 f f f f t f v u 1 0 2278 "19" _null_ _null_ _null_ _null_ _null_ pg_drop_replication_slot _null_ _null_ _null_ )); DESCR("drop a replication slot"); -DATA(insert OID = 3781 ( pg_get_replication_slots PGNSP PGUID 12 1 10 0 0 f f f f f t s s 0 0 2249 "" "{19,19,25,26,16,16,23,28,28,3220,3220}" "{o,o,o,o,o,o,o,o,o,o,o}" "{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn}" _null_ _null_ pg_get_replication_slots _null_ _null_ _null_ )); +DATA(insert OID = 3781 ( pg_get_replication_slots PGNSP PGUID 12 1 10 0 0 f f f f f t s s 0 0 2249 "" "{19,19,25,26,16,16,23,28,28,3220,3220,16,3220}" "{o,o,o,o,o,o,o,o,o,o,o,o,o}" "{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn,live,distance}" _null_ _null_ pg_get_replication_slots _null_ _null_ _null_ )); DESCR("information about replication slots currently in use"); DATA(insert OID = 3786 ( pg_create_logical_replication_slot PGNSP PGUID 12 1 0 0 0 f f f f t f v u 3 0 2249 "19 19 16" "{19,19,16,25,3220}" "{i,i,i,o,o}" "{slot_name,plugin,temporary,slot_name,lsn}" _null_ _null_ pg_create_logical_replication_slot _null_ _null_ _null_ )); DESCR("set up a logical replication slot"); diff --git a/src/test/regress/expected/rules.out b/src/test/regress/expected/rules.out index f1c1b44..16a99d8 100644 --- a/src/test/regress/expected/rules.out +++ b/src/test/regress/expected/rules.out @@ -1451,8 +1451,10 @@ pg_replication_slots| SELECT l.slot_name, l.xmin, l.catalog_xmin, l.restart_lsn, - l.confirmed_flush_lsn - FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn) + l.confirmed_flush_lsn, + l.live, + l.distance + FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn, live, distance) LEFT JOIN pg_database d ON ((l.datoid = d.oid))); pg_roles| SELECT pg_authid.rolname, pg_authid.rolsuper, -- 2.9.2 ----Next_Part(Tue_Oct_31_18_43_10_2017_912)-- Content-Type: text/plain Content-Disposition: inline Content-Transfer-Encoding: 8bit MIME-Version: 1.0 -- Sent via pgsql-hackers mailing list ([email protected]) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers ----Next_Part(Tue_Oct_31_18_43_10_2017_912)---- ^ permalink raw reply [nested|flat] 13+ messages in thread
* [PATCH 2/3] Add monitoring aid for max_replication_slots. @ 2017-12-21 12:23 Kyotaro Horiguchi <[email protected]> 0 siblings, 0 replies; 13+ messages in thread From: Kyotaro Horiguchi @ 2017-12-21 12:23 UTC (permalink / raw) Adds two columns "status" and "min_secure_lsn" in pg_replication_slot. Setting max_slot_wal_keep_size, long-disconnected slots may lose sync. The two columns shows that a slot can be reconnected or not, or about to lose required WAL segments. And the LSN back to where the next checkpoint will secure. --- src/backend/access/transam/xlog.c | 93 ++++++++++++++++++++++++++++++++++++ src/backend/catalog/system_views.sql | 4 +- src/backend/replication/slotfuncs.c | 25 +++++++++- src/include/access/xlog.h | 1 + src/include/catalog/pg_proc.h | 2 +- src/test/regress/expected/rules.out | 6 ++- 6 files changed, 126 insertions(+), 5 deletions(-) diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 723a983..b630224 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9349,6 +9349,99 @@ CreateRestartPoint(int flags) return true; } + +/* + * Returns the segment number of the oldest file in XLOG directory. + */ +static XLogSegNo +GetOldestXLogFileSegNo(void) +{ + DIR *xldir; + struct dirent *xlde; + XLogSegNo segno = 0; + + xldir = AllocateDir(XLOGDIR); + if (xldir == NULL) + ereport(ERROR, + (errcode_for_file_access(), + errmsg("could not open write-ahead log directory \"%s\": %m", + XLOGDIR))); + + while ((xlde = ReadDir(xldir, XLOGDIR)) != NULL) + { + TimeLineID tli; + XLogSegNo fsegno; + + /* Ignore files that are not XLOG segments */ + if (!IsXLogFileName(xlde->d_name) && + !IsPartialXLogFileName(xlde->d_name)) + continue; + + XLogFromFileName(xlde->d_name, &tli, &fsegno, wal_segment_size); + + /* get minimum segment ignorig timeline ID */ + if (segno == 0 || fsegno < segno) + segno = fsegno; + } + + return segno; +} + +/* + * Check if the record on the given restartLSN is present in XLOG files. + * + * Returns true if it is present. If minSecureLSN is given, it receives the + * LSN at the beginning of the oldest existing WAL segment. + */ +bool +IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minSecureLSN) +{ + XLogRecPtr currpos; + XLogSegNo restartSeg; + XLogSegNo tailSeg; + XLogSegNo oldestSeg; + + Assert(!XLogRecPtrIsInvalid(restartLSN)); + + currpos = GetXLogWriteRecPtr(); + + SpinLockAcquire(&XLogCtl->info_lck); + oldestSeg = XLogCtl->lastRemovedSegNo; + SpinLockRelease(&XLogCtl->info_lck); + + /* + * oldestSeg is zero before at least one segment has been removed since + * startup. Use oldest segno taken from file names. + */ + if (oldestSeg == 0) + { + static XLogSegNo oldestFileSeg = 0; + + if (oldestFileSeg == 0) + oldestFileSeg = GetOldestXLogFileSegNo(); + /* let it have the same meaning with lastRemovedSegNo here */ + oldestSeg = oldestFileSeg - 1; + } + + /* oldest segment is just after the last removed segment */ + oldestSeg++; + + XLByteToSeg(restartLSN, restartSeg, wal_segment_size); + + + if (minSecureLSN) + { + XLogRecPtr slotPtr = XLogGetReplicationSlotMinimumLSN(); + Assert(!XLogRecPtrIsInvalid(slotPtr)); + + tailSeg = GetMinSecuredSegment(currpos, slotPtr); + + XLogSegNoOffsetToRecPtr(tailSeg, 0, *minSecureLSN, wal_segment_size); + } + + return oldestSeg <= restartSeg; +} + /* * Returns minimum segment number the next checktpoint must leave considering * wal_keep_segments, replication slots and max_slot_wal_keep_size. diff --git a/src/backend/catalog/system_views.sql b/src/backend/catalog/system_views.sql index 394aea8..4167146 100644 --- a/src/backend/catalog/system_views.sql +++ b/src/backend/catalog/system_views.sql @@ -793,7 +793,9 @@ CREATE VIEW pg_replication_slots AS L.xmin, L.catalog_xmin, L.restart_lsn, - L.confirmed_flush_lsn + L.confirmed_flush_lsn, + L.status, + L.min_secure_lsn FROM pg_get_replication_slots() AS L LEFT JOIN pg_database D ON (L.datoid = D.oid); diff --git a/src/backend/replication/slotfuncs.c b/src/backend/replication/slotfuncs.c index ab776e8..3880807 100644 --- a/src/backend/replication/slotfuncs.c +++ b/src/backend/replication/slotfuncs.c @@ -182,7 +182,7 @@ pg_drop_replication_slot(PG_FUNCTION_ARGS) Datum pg_get_replication_slots(PG_FUNCTION_ARGS) { -#define PG_GET_REPLICATION_SLOTS_COLS 11 +#define PG_GET_REPLICATION_SLOTS_COLS 13 ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; TupleDesc tupdesc; Tuplestorestate *tupstore; @@ -304,6 +304,29 @@ pg_get_replication_slots(PG_FUNCTION_ARGS) else nulls[i++] = true; + if (restart_lsn == InvalidXLogRecPtr) + { + values[i++] = CStringGetTextDatum("unknown"); + values[i++] = LSNGetDatum(InvalidXLogRecPtr); + } + else + { + XLogRecPtr min_secure_lsn; + char *status = "broken"; + + if (BoolGetDatum(IsLsnStillAvaiable(restart_lsn, + &min_secure_lsn))) + { + if (min_secure_lsn <= restart_lsn) + status = "secured"; + else + status = "insecured"; + } + + values[i++] = CStringGetTextDatum(status); + values[i++] = LSNGetDatum(min_secure_lsn); + } + tuplestore_putvalues(tupstore, tupdesc, values, nulls); } LWLockRelease(ReplicationSlotControlLock); diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h index 45eb51a..542df28 100644 --- a/src/include/access/xlog.h +++ b/src/include/access/xlog.h @@ -269,6 +269,7 @@ extern void ShutdownXLOG(int code, Datum arg); extern void InitXLOGAccess(void); extern void CreateCheckPoint(int flags); extern bool CreateRestartPoint(int flags); +extern bool IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minSecureLSN); extern void XLogPutNextOid(Oid nextOid); extern XLogRecPtr XLogRestorePoint(const char *rpName); extern void UpdateFullPageWrites(void); diff --git a/src/include/catalog/pg_proc.h b/src/include/catalog/pg_proc.h index c969375..1157438 100644 --- a/src/include/catalog/pg_proc.h +++ b/src/include/catalog/pg_proc.h @@ -5340,7 +5340,7 @@ DATA(insert OID = 3779 ( pg_create_physical_replication_slot PGNSP PGUID 12 1 0 DESCR("create a physical replication slot"); DATA(insert OID = 3780 ( pg_drop_replication_slot PGNSP PGUID 12 1 0 0 0 f f f f t f v u 1 0 2278 "19" _null_ _null_ _null_ _null_ _null_ pg_drop_replication_slot _null_ _null_ _null_ )); DESCR("drop a replication slot"); -DATA(insert OID = 3781 ( pg_get_replication_slots PGNSP PGUID 12 1 10 0 0 f f f f f t s s 0 0 2249 "" "{19,19,25,26,16,16,23,28,28,3220,3220}" "{o,o,o,o,o,o,o,o,o,o,o}" "{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn}" _null_ _null_ pg_get_replication_slots _null_ _null_ _null_ )); +DATA(insert OID = 3781 ( pg_get_replication_slots PGNSP PGUID 12 1 10 0 0 f f f f f t s s 0 0 2249 "" "{19,19,25,26,16,16,23,28,28,3220,3220,25,3220}" "{o,o,o,o,o,o,o,o,o,o,o,o,o}" "{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn,status,min_secure_lsn}" _null_ _null_ pg_get_replication_slots _null_ _null_ _null_ )); DESCR("information about replication slots currently in use"); DATA(insert OID = 3786 ( pg_create_logical_replication_slot PGNSP PGUID 12 1 0 0 0 f f f f t f v u 3 0 2249 "19 19 16" "{19,19,16,25,3220}" "{i,i,i,o,o}" "{slot_name,plugin,temporary,slot_name,lsn}" _null_ _null_ pg_create_logical_replication_slot _null_ _null_ _null_ )); DESCR("set up a logical replication slot"); diff --git a/src/test/regress/expected/rules.out b/src/test/regress/expected/rules.out index f1c1b44..d9d74a3 100644 --- a/src/test/regress/expected/rules.out +++ b/src/test/regress/expected/rules.out @@ -1451,8 +1451,10 @@ pg_replication_slots| SELECT l.slot_name, l.xmin, l.catalog_xmin, l.restart_lsn, - l.confirmed_flush_lsn - FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn) + l.confirmed_flush_lsn, + l.status, + l.min_secure_lsn + FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn, status, min_secure_lsn) LEFT JOIN pg_database d ON ((l.datoid = d.oid))); pg_roles| SELECT pg_authid.rolname, pg_authid.rolsuper, -- 2.9.2 ----Next_Part(Fri_Dec_22_15_03_20_2017_403)-- Content-Type: Text/X-Patch; charset=us-ascii Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="0003-TAP-test-for-the-slot-limit-feature.patch" ^ permalink raw reply [nested|flat] 13+ messages in thread
* [PATCH 2/4] Add monitoring aid for max_replication_slots. @ 2017-12-21 12:23 Kyotaro Horiguchi <[email protected]> 0 siblings, 0 replies; 13+ messages in thread From: Kyotaro Horiguchi @ 2017-12-21 12:23 UTC (permalink / raw) Adds two columns "status" and "min_secure_lsn" in pg_replication_slot. Setting max_slot_wal_keep_size, long-disconnected slots may lose sync. The two columns shows that a slot can be reconnected or not, or about to lose required WAL segments. And the LSN back to where the next checkpoint will secure. --- src/backend/access/transam/xlog.c | 93 ++++++++++++++++++++++++++++++++++++ src/backend/catalog/system_views.sql | 4 +- src/backend/replication/slotfuncs.c | 25 +++++++++- src/include/access/xlog.h | 1 + src/include/catalog/pg_proc.h | 2 +- src/test/regress/expected/rules.out | 6 ++- 6 files changed, 126 insertions(+), 5 deletions(-) diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index bdb7156..a8423f7 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9349,6 +9349,99 @@ CreateRestartPoint(int flags) return true; } + +/* + * Returns the segment number of the oldest file in XLOG directory. + */ +static XLogSegNo +GetOldestXLogFileSegNo(void) +{ + DIR *xldir; + struct dirent *xlde; + XLogSegNo segno = 0; + + xldir = AllocateDir(XLOGDIR); + if (xldir == NULL) + ereport(ERROR, + (errcode_for_file_access(), + errmsg("could not open write-ahead log directory \"%s\": %m", + XLOGDIR))); + + while ((xlde = ReadDir(xldir, XLOGDIR)) != NULL) + { + TimeLineID tli; + XLogSegNo fsegno; + + /* Ignore files that are not XLOG segments */ + if (!IsXLogFileName(xlde->d_name) && + !IsPartialXLogFileName(xlde->d_name)) + continue; + + XLogFromFileName(xlde->d_name, &tli, &fsegno, wal_segment_size); + + /* get minimum segment ignorig timeline ID */ + if (segno == 0 || fsegno < segno) + segno = fsegno; + } + + return segno; +} + +/* + * Check if the record on the given restartLSN is present in XLOG files. + * + * Returns true if it is present. If minKeepLSN is given, it receives the + * LSN at the beginning of the oldest existing WAL segment. + */ +bool +IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minKeepLSN) +{ + XLogRecPtr currpos; + XLogSegNo restartSeg; + XLogSegNo tailSeg; + XLogSegNo oldestSeg; + + Assert(!XLogRecPtrIsInvalid(restartLSN)); + + currpos = GetXLogWriteRecPtr(); + + SpinLockAcquire(&XLogCtl->info_lck); + oldestSeg = XLogCtl->lastRemovedSegNo; + SpinLockRelease(&XLogCtl->info_lck); + + /* + * oldestSeg is zero before at least one segment has been removed since + * startup. Use oldest segno taken from file names. + */ + if (oldestSeg == 0) + { + static XLogSegNo oldestFileSeg = 0; + + if (oldestFileSeg == 0) + oldestFileSeg = GetOldestXLogFileSegNo(); + /* let it have the same meaning with lastRemovedSegNo here */ + oldestSeg = oldestFileSeg - 1; + } + + /* oldest segment is just after the last removed segment */ + oldestSeg++; + + XLByteToSeg(restartLSN, restartSeg, wal_segment_size); + + + if (minKeepLSN) + { + XLogRecPtr slotPtr = XLogGetReplicationSlotMinimumLSN(); + Assert(!XLogRecPtrIsInvalid(slotPtr)); + + tailSeg = GetMinKeepSegment(currpos, slotPtr); + + XLogSegNoOffsetToRecPtr(tailSeg, 0, *minKeepLSN, wal_segment_size); + } + + return oldestSeg <= restartSeg; +} + /* * Returns minimum segment number the next checktpoint must leave considering * wal_keep_segments, replication slots and max_slot_wal_keep_size. diff --git a/src/backend/catalog/system_views.sql b/src/backend/catalog/system_views.sql index 5652e9e..cd714cc 100644 --- a/src/backend/catalog/system_views.sql +++ b/src/backend/catalog/system_views.sql @@ -793,7 +793,9 @@ CREATE VIEW pg_replication_slots AS L.xmin, L.catalog_xmin, L.restart_lsn, - L.confirmed_flush_lsn + L.confirmed_flush_lsn, + L.wal_status, + L.min_keep_lsn FROM pg_get_replication_slots() AS L LEFT JOIN pg_database D ON (L.datoid = D.oid); diff --git a/src/backend/replication/slotfuncs.c b/src/backend/replication/slotfuncs.c index b02df59..84f4154 100644 --- a/src/backend/replication/slotfuncs.c +++ b/src/backend/replication/slotfuncs.c @@ -182,7 +182,7 @@ pg_drop_replication_slot(PG_FUNCTION_ARGS) Datum pg_get_replication_slots(PG_FUNCTION_ARGS) { -#define PG_GET_REPLICATION_SLOTS_COLS 11 +#define PG_GET_REPLICATION_SLOTS_COLS 13 ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; TupleDesc tupdesc; Tuplestorestate *tupstore; @@ -304,6 +304,29 @@ pg_get_replication_slots(PG_FUNCTION_ARGS) else nulls[i++] = true; + if (restart_lsn == InvalidXLogRecPtr) + { + values[i++] = CStringGetTextDatum("unknown"); + values[i++] = LSNGetDatum(InvalidXLogRecPtr); + } + else + { + XLogRecPtr min_keep_lsn; + char *status = "lost"; + + if (BoolGetDatum(IsLsnStillAvaiable(restart_lsn, + &min_keep_lsn))) + { + if (min_keep_lsn <= restart_lsn) + status = "streaming"; + else + status = "keeping"; + } + + values[i++] = CStringGetTextDatum(status); + values[i++] = LSNGetDatum(min_keep_lsn); + } + tuplestore_putvalues(tupstore, tupdesc, values, nulls); } LWLockRelease(ReplicationSlotControlLock); diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h index 12cd0d1..52e64f3 100644 --- a/src/include/access/xlog.h +++ b/src/include/access/xlog.h @@ -269,6 +269,7 @@ extern void ShutdownXLOG(int code, Datum arg); extern void InitXLOGAccess(void); extern void CreateCheckPoint(int flags); extern bool CreateRestartPoint(int flags); +extern bool IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minSecureLSN); extern void XLogPutNextOid(Oid nextOid); extern XLogRecPtr XLogRestorePoint(const char *rpName); extern void UpdateFullPageWrites(void); diff --git a/src/include/catalog/pg_proc.h b/src/include/catalog/pg_proc.h index 298e0ae..93f62d7 100644 --- a/src/include/catalog/pg_proc.h +++ b/src/include/catalog/pg_proc.h @@ -5345,7 +5345,7 @@ DATA(insert OID = 3779 ( pg_create_physical_replication_slot PGNSP PGUID 12 1 0 DESCR("create a physical replication slot"); DATA(insert OID = 3780 ( pg_drop_replication_slot PGNSP PGUID 12 1 0 0 0 f f f f t f v u 1 0 2278 "19" _null_ _null_ _null_ _null_ _null_ pg_drop_replication_slot _null_ _null_ _null_ )); DESCR("drop a replication slot"); -DATA(insert OID = 3781 ( pg_get_replication_slots PGNSP PGUID 12 1 10 0 0 f f f f f t s s 0 0 2249 "" "{19,19,25,26,16,16,23,28,28,3220,3220}" "{o,o,o,o,o,o,o,o,o,o,o}" "{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn}" _null_ _null_ pg_get_replication_slots _null_ _null_ _null_ )); +DATA(insert OID = 3781 ( pg_get_replication_slots PGNSP PGUID 12 1 10 0 0 f f f f f t s s 0 0 2249 "" "{19,19,25,26,16,16,23,28,28,3220,3220,25,3220}" "{o,o,o,o,o,o,o,o,o,o,o,o,o}" "{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn,wal_status,min_keep_lsn}" _null_ _null_ pg_get_replication_slots _null_ _null_ _null_ )); DESCR("information about replication slots currently in use"); DATA(insert OID = 3786 ( pg_create_logical_replication_slot PGNSP PGUID 12 1 0 0 0 f f f f t f v u 3 0 2249 "19 19 16" "{19,19,16,25,3220}" "{i,i,i,o,o}" "{slot_name,plugin,temporary,slot_name,lsn}" _null_ _null_ pg_create_logical_replication_slot _null_ _null_ _null_ )); DESCR("set up a logical replication slot"); diff --git a/src/test/regress/expected/rules.out b/src/test/regress/expected/rules.out index f1c1b44..75d44af 100644 --- a/src/test/regress/expected/rules.out +++ b/src/test/regress/expected/rules.out @@ -1451,8 +1451,10 @@ pg_replication_slots| SELECT l.slot_name, l.xmin, l.catalog_xmin, l.restart_lsn, - l.confirmed_flush_lsn - FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn) + l.confirmed_flush_lsn, + l.wal_status, + l.min_keep_lsn + FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn, wal_status, min_keep_lsn) LEFT JOIN pg_database d ON ((l.datoid = d.oid))); pg_roles| SELECT pg_authid.rolname, pg_authid.rolsuper, -- 2.9.2 ----Next_Part(Thu_Jan_11_15_59_10_2018_946)-- Content-Type: Text/X-Patch; charset=us-ascii Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="0003-TAP-test-for-the-slot-limit-feature.patch" ^ permalink raw reply [nested|flat] 13+ messages in thread
* [PATCH 2/4] Add monitoring aid for max_replication_slots. @ 2017-12-21 12:23 Kyotaro Horiguchi <[email protected]> 0 siblings, 0 replies; 13+ messages in thread From: Kyotaro Horiguchi @ 2017-12-21 12:23 UTC (permalink / raw) Adds two columns "status" and "min_secure_lsn" in pg_replication_slot. Setting max_slot_wal_keep_size, long-disconnected slots may lose sync. The two columns shows that a slot can be reconnected or not, or about to lose required WAL segments. And the LSN back to where the next checkpoint will secure. --- contrib/test_decoding/expected/ddl.out | 4 +- src/backend/access/transam/xlog.c | 95 ++++++++++++++++++++++++++++++++++ src/backend/catalog/system_views.sql | 4 +- src/backend/replication/slotfuncs.c | 25 ++++++++- src/include/access/xlog.h | 1 + src/include/catalog/pg_proc.dat | 6 +-- src/test/regress/expected/rules.out | 6 ++- 7 files changed, 132 insertions(+), 9 deletions(-) diff --git a/contrib/test_decoding/expected/ddl.out b/contrib/test_decoding/expected/ddl.out index b7c76469fc..276f7f6efd 100644 --- a/contrib/test_decoding/expected/ddl.out +++ b/contrib/test_decoding/expected/ddl.out @@ -706,7 +706,7 @@ SELECT pg_drop_replication_slot('regression_slot'); /* check that the slot is gone */ SELECT * FROM pg_replication_slots; - slot_name | plugin | slot_type | datoid | database | temporary | active | active_pid | xmin | catalog_xmin | restart_lsn | confirmed_flush_lsn ------------+--------+-----------+--------+----------+-----------+--------+------------+------+--------------+-------------+--------------------- + slot_name | plugin | slot_type | datoid | database | temporary | active | active_pid | xmin | catalog_xmin | restart_lsn | confirmed_flush_lsn | wal_status | min_keep_lsn +-----------+--------+-----------+--------+----------+-----------+--------+------------+------+--------------+-------------+---------------------+------------+-------------- (0 rows) diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 959d18e029..5c16750e89 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9379,6 +9379,101 @@ CreateRestartPoint(int flags) return true; } + +/* + * Returns the segment number of the oldest file in XLOG directory. + */ +static XLogSegNo +GetOldestXLogFileSegNo(void) +{ + DIR *xldir; + struct dirent *xlde; + XLogSegNo segno = 0; + + xldir = AllocateDir(XLOGDIR); + if (xldir == NULL) + ereport(ERROR, + (errcode_for_file_access(), + errmsg("could not open write-ahead log directory \"%s\": %m", + XLOGDIR))); + + while ((xlde = ReadDir(xldir, XLOGDIR)) != NULL) + { + TimeLineID tli; + XLogSegNo fsegno; + + /* Ignore files that are not XLOG segments */ + if (!IsXLogFileName(xlde->d_name) && + !IsPartialXLogFileName(xlde->d_name)) + continue; + + XLogFromFileName(xlde->d_name, &tli, &fsegno, wal_segment_size); + + /* get minimum segment ignorig timeline ID */ + if (segno == 0 || fsegno < segno) + segno = fsegno; + } + + FreeDir(xldir); + + return segno; +} + +/* + * Check if the record on the given restartLSN is present in XLOG files. + * + * Returns true if it is present. If minKeepLSN is given, it receives the + * LSN at the beginning of the oldest existing WAL segment. + */ +bool +IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minKeepLSN) +{ + XLogRecPtr currpos; + XLogSegNo restartSeg; + XLogSegNo tailSeg; + XLogSegNo oldestSeg; + + Assert(!XLogRecPtrIsInvalid(restartLSN)); + + currpos = GetXLogWriteRecPtr(); + + SpinLockAcquire(&XLogCtl->info_lck); + oldestSeg = XLogCtl->lastRemovedSegNo; + SpinLockRelease(&XLogCtl->info_lck); + + /* + * oldestSeg is zero before at least one segment has been removed since + * startup. Use oldest segno taken from file names. + */ + if (oldestSeg == 0) + { + static XLogSegNo oldestFileSeg = 0; + + if (oldestFileSeg == 0) + oldestFileSeg = GetOldestXLogFileSegNo(); + /* let it have the same meaning with lastRemovedSegNo here */ + oldestSeg = oldestFileSeg - 1; + } + + /* oldest segment is just after the last removed segment */ + oldestSeg++; + + XLByteToSeg(restartLSN, restartSeg, wal_segment_size); + + + if (minKeepLSN) + { + XLogRecPtr slotPtr = XLogGetReplicationSlotMinimumLSN(); + Assert(!XLogRecPtrIsInvalid(slotPtr)); + + tailSeg = GetOldestKeepSegment(currpos, slotPtr); + + XLogSegNoOffsetToRecPtr(tailSeg, 0, *minKeepLSN, wal_segment_size); + } + + return oldestSeg <= restartSeg; +} + /* * Returns minimum segment number the next checktpoint must leave considering * wal_keep_segments, replication slots and max_slot_wal_keep_size. diff --git a/src/backend/catalog/system_views.sql b/src/backend/catalog/system_views.sql index 8cd8bf40ac..1664a086e9 100644 --- a/src/backend/catalog/system_views.sql +++ b/src/backend/catalog/system_views.sql @@ -797,7 +797,9 @@ CREATE VIEW pg_replication_slots AS L.xmin, L.catalog_xmin, L.restart_lsn, - L.confirmed_flush_lsn + L.confirmed_flush_lsn, + L.wal_status, + L.min_keep_lsn FROM pg_get_replication_slots() AS L LEFT JOIN pg_database D ON (L.datoid = D.oid); diff --git a/src/backend/replication/slotfuncs.c b/src/backend/replication/slotfuncs.c index 2806e1076c..f13aa4d455 100644 --- a/src/backend/replication/slotfuncs.c +++ b/src/backend/replication/slotfuncs.c @@ -185,7 +185,7 @@ pg_drop_replication_slot(PG_FUNCTION_ARGS) Datum pg_get_replication_slots(PG_FUNCTION_ARGS) { -#define PG_GET_REPLICATION_SLOTS_COLS 11 +#define PG_GET_REPLICATION_SLOTS_COLS 13 ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; TupleDesc tupdesc; Tuplestorestate *tupstore; @@ -307,6 +307,29 @@ pg_get_replication_slots(PG_FUNCTION_ARGS) else nulls[i++] = true; + if (restart_lsn == InvalidXLogRecPtr) + { + values[i++] = CStringGetTextDatum("unknown"); + values[i++] = LSNGetDatum(InvalidXLogRecPtr); + } + else + { + XLogRecPtr min_keep_lsn; + char *status = "lost"; + + if (BoolGetDatum(IsLsnStillAvaiable(restart_lsn, + &min_keep_lsn))) + { + if (min_keep_lsn <= restart_lsn) + status = "streaming"; + else + status = "keeping"; + } + + values[i++] = CStringGetTextDatum(status); + values[i++] = LSNGetDatum(min_keep_lsn); + } + tuplestore_putvalues(tupstore, tupdesc, values, nulls); } LWLockRelease(ReplicationSlotControlLock); diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h index 12cd0d1d10..52e64f392d 100644 --- a/src/include/access/xlog.h +++ b/src/include/access/xlog.h @@ -269,6 +269,7 @@ extern void ShutdownXLOG(int code, Datum arg); extern void InitXLOGAccess(void); extern void CreateCheckPoint(int flags); extern bool CreateRestartPoint(int flags); +extern bool IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minSecureLSN); extern void XLogPutNextOid(Oid nextOid); extern XLogRecPtr XLogRestorePoint(const char *rpName); extern void UpdateFullPageWrites(void); diff --git a/src/include/catalog/pg_proc.dat b/src/include/catalog/pg_proc.dat index 40d54ed030..ef8f0eab91 100644 --- a/src/include/catalog/pg_proc.dat +++ b/src/include/catalog/pg_proc.dat @@ -9796,9 +9796,9 @@ proname => 'pg_get_replication_slots', prorows => '10', proisstrict => 'f', proretset => 't', provolatile => 's', prorettype => 'record', proargtypes => '', - proallargtypes => '{name,name,text,oid,bool,bool,int4,xid,xid,pg_lsn,pg_lsn}', - proargmodes => '{o,o,o,o,o,o,o,o,o,o,o}', - proargnames => '{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn}', + proallargtypes => '{name,name,text,oid,bool,bool,int4,xid,xid,pg_lsn,pg_lsn,text,pg_lsn}', + proargmodes => '{o,o,o,o,o,o,o,o,o,o,o,o,o}', + proargnames => '{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn,wal_status,min_keep_lsn}', prosrc => 'pg_get_replication_slots' }, { oid => '3786', descr => 'set up a logical replication slot', proname => 'pg_create_logical_replication_slot', provolatile => 'v', diff --git a/src/test/regress/expected/rules.out b/src/test/regress/expected/rules.out index ae0cd253d5..93f6bff77e 100644 --- a/src/test/regress/expected/rules.out +++ b/src/test/regress/expected/rules.out @@ -1451,8 +1451,10 @@ pg_replication_slots| SELECT l.slot_name, l.xmin, l.catalog_xmin, l.restart_lsn, - l.confirmed_flush_lsn - FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn) + l.confirmed_flush_lsn, + l.wal_status, + l.min_keep_lsn + FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn, wal_status, min_keep_lsn) LEFT JOIN pg_database d ON ((l.datoid = d.oid))); pg_roles| SELECT pg_authid.rolname, pg_authid.rolsuper, -- 2.16.3 ----Next_Part(Wed_Jul_04_17_28_38_2018_173)-- Content-Type: Text/X-Patch; charset=us-ascii Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="v3-0003-TAP-test-for-the-slot-limit-feature.patch" ^ permalink raw reply [nested|flat] 13+ messages in thread
* [PATCH 2/4] Add monitoring aid for max_replication_slots. @ 2017-12-21 12:23 Kyotaro Horiguchi <[email protected]> 0 siblings, 0 replies; 13+ messages in thread From: Kyotaro Horiguchi @ 2017-12-21 12:23 UTC (permalink / raw) Adds two columns "status" and "min_secure_lsn" in pg_replication_slot. Setting max_slot_wal_keep_size, long-disconnected slots may lose sync. The two columns shows that a slot can be reconnected or not, or about to lose required WAL segments. And the LSN back to where the next checkpoint will secure. --- contrib/test_decoding/expected/ddl.out | 4 +- src/backend/access/transam/xlog.c | 93 ++++++++++++++++++++++++++++++++++ src/backend/catalog/system_views.sql | 4 +- src/backend/replication/slotfuncs.c | 25 ++++++++- src/include/access/xlog.h | 1 + src/include/catalog/pg_proc.h | 2 +- src/test/regress/expected/rules.out | 6 ++- 7 files changed, 128 insertions(+), 7 deletions(-) diff --git a/contrib/test_decoding/expected/ddl.out b/contrib/test_decoding/expected/ddl.out index 1e22c1e..92cd56a 100644 --- a/contrib/test_decoding/expected/ddl.out +++ b/contrib/test_decoding/expected/ddl.out @@ -702,7 +702,7 @@ SELECT pg_drop_replication_slot('regression_slot'); /* check that the slot is gone */ SELECT * FROM pg_replication_slots; - slot_name | plugin | slot_type | datoid | database | temporary | active | active_pid | xmin | catalog_xmin | restart_lsn | confirmed_flush_lsn ------------+--------+-----------+--------+----------+-----------+--------+------------+------+--------------+-------------+--------------------- + slot_name | plugin | slot_type | datoid | database | temporary | active | active_pid | xmin | catalog_xmin | restart_lsn | confirmed_flush_lsn | wal_status | min_keep_lsn +-----------+--------+-----------+--------+----------+-----------+--------+------------+------+--------------+-------------+---------------------+------------+-------------- (0 rows) diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 55694cb..cbece1b 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9349,6 +9349,99 @@ CreateRestartPoint(int flags) return true; } + +/* + * Returns the segment number of the oldest file in XLOG directory. + */ +static XLogSegNo +GetOldestXLogFileSegNo(void) +{ + DIR *xldir; + struct dirent *xlde; + XLogSegNo segno = 0; + + xldir = AllocateDir(XLOGDIR); + if (xldir == NULL) + ereport(ERROR, + (errcode_for_file_access(), + errmsg("could not open write-ahead log directory \"%s\": %m", + XLOGDIR))); + + while ((xlde = ReadDir(xldir, XLOGDIR)) != NULL) + { + TimeLineID tli; + XLogSegNo fsegno; + + /* Ignore files that are not XLOG segments */ + if (!IsXLogFileName(xlde->d_name) && + !IsPartialXLogFileName(xlde->d_name)) + continue; + + XLogFromFileName(xlde->d_name, &tli, &fsegno, wal_segment_size); + + /* get minimum segment ignorig timeline ID */ + if (segno == 0 || fsegno < segno) + segno = fsegno; + } + + return segno; +} + +/* + * Check if the record on the given restartLSN is present in XLOG files. + * + * Returns true if it is present. If minKeepLSN is given, it receives the + * LSN at the beginning of the oldest existing WAL segment. + */ +bool +IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minKeepLSN) +{ + XLogRecPtr currpos; + XLogSegNo restartSeg; + XLogSegNo tailSeg; + XLogSegNo oldestSeg; + + Assert(!XLogRecPtrIsInvalid(restartLSN)); + + currpos = GetXLogWriteRecPtr(); + + SpinLockAcquire(&XLogCtl->info_lck); + oldestSeg = XLogCtl->lastRemovedSegNo; + SpinLockRelease(&XLogCtl->info_lck); + + /* + * oldestSeg is zero before at least one segment has been removed since + * startup. Use oldest segno taken from file names. + */ + if (oldestSeg == 0) + { + static XLogSegNo oldestFileSeg = 0; + + if (oldestFileSeg == 0) + oldestFileSeg = GetOldestXLogFileSegNo(); + /* let it have the same meaning with lastRemovedSegNo here */ + oldestSeg = oldestFileSeg - 1; + } + + /* oldest segment is just after the last removed segment */ + oldestSeg++; + + XLByteToSeg(restartLSN, restartSeg, wal_segment_size); + + + if (minKeepLSN) + { + XLogRecPtr slotPtr = XLogGetReplicationSlotMinimumLSN(); + Assert(!XLogRecPtrIsInvalid(slotPtr)); + + tailSeg = GetMinKeepSegment(currpos, slotPtr); + + XLogSegNoOffsetToRecPtr(tailSeg, 0, *minKeepLSN, wal_segment_size); + } + + return oldestSeg <= restartSeg; +} + /* * Returns minimum segment number the next checktpoint must leave considering * wal_keep_segments, replication slots and max_slot_wal_keep_size. diff --git a/src/backend/catalog/system_views.sql b/src/backend/catalog/system_views.sql index 5652e9e..cd714cc 100644 --- a/src/backend/catalog/system_views.sql +++ b/src/backend/catalog/system_views.sql @@ -793,7 +793,9 @@ CREATE VIEW pg_replication_slots AS L.xmin, L.catalog_xmin, L.restart_lsn, - L.confirmed_flush_lsn + L.confirmed_flush_lsn, + L.wal_status, + L.min_keep_lsn FROM pg_get_replication_slots() AS L LEFT JOIN pg_database D ON (L.datoid = D.oid); diff --git a/src/backend/replication/slotfuncs.c b/src/backend/replication/slotfuncs.c index cf2195b..e6e0386 100644 --- a/src/backend/replication/slotfuncs.c +++ b/src/backend/replication/slotfuncs.c @@ -185,7 +185,7 @@ pg_drop_replication_slot(PG_FUNCTION_ARGS) Datum pg_get_replication_slots(PG_FUNCTION_ARGS) { -#define PG_GET_REPLICATION_SLOTS_COLS 11 +#define PG_GET_REPLICATION_SLOTS_COLS 13 ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; TupleDesc tupdesc; Tuplestorestate *tupstore; @@ -307,6 +307,29 @@ pg_get_replication_slots(PG_FUNCTION_ARGS) else nulls[i++] = true; + if (restart_lsn == InvalidXLogRecPtr) + { + values[i++] = CStringGetTextDatum("unknown"); + values[i++] = LSNGetDatum(InvalidXLogRecPtr); + } + else + { + XLogRecPtr min_keep_lsn; + char *status = "lost"; + + if (BoolGetDatum(IsLsnStillAvaiable(restart_lsn, + &min_keep_lsn))) + { + if (min_keep_lsn <= restart_lsn) + status = "streaming"; + else + status = "keeping"; + } + + values[i++] = CStringGetTextDatum(status); + values[i++] = LSNGetDatum(min_keep_lsn); + } + tuplestore_putvalues(tupstore, tupdesc, values, nulls); } LWLockRelease(ReplicationSlotControlLock); diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h index 12cd0d1..52e64f3 100644 --- a/src/include/access/xlog.h +++ b/src/include/access/xlog.h @@ -269,6 +269,7 @@ extern void ShutdownXLOG(int code, Datum arg); extern void InitXLOGAccess(void); extern void CreateCheckPoint(int flags); extern bool CreateRestartPoint(int flags); +extern bool IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minSecureLSN); extern void XLogPutNextOid(Oid nextOid); extern XLogRecPtr XLogRestorePoint(const char *rpName); extern void UpdateFullPageWrites(void); diff --git a/src/include/catalog/pg_proc.h b/src/include/catalog/pg_proc.h index f01648c..0712e60 100644 --- a/src/include/catalog/pg_proc.h +++ b/src/include/catalog/pg_proc.h @@ -5345,7 +5345,7 @@ DATA(insert OID = 3779 ( pg_create_physical_replication_slot PGNSP PGUID 12 1 0 DESCR("create a physical replication slot"); DATA(insert OID = 3780 ( pg_drop_replication_slot PGNSP PGUID 12 1 0 0 0 f f f f t f v u 1 0 2278 "19" _null_ _null_ _null_ _null_ _null_ pg_drop_replication_slot _null_ _null_ _null_ )); DESCR("drop a replication slot"); -DATA(insert OID = 3781 ( pg_get_replication_slots PGNSP PGUID 12 1 10 0 0 f f f f f t s s 0 0 2249 "" "{19,19,25,26,16,16,23,28,28,3220,3220}" "{o,o,o,o,o,o,o,o,o,o,o}" "{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn}" _null_ _null_ pg_get_replication_slots _null_ _null_ _null_ )); +DATA(insert OID = 3781 ( pg_get_replication_slots PGNSP PGUID 12 1 10 0 0 f f f f f t s s 0 0 2249 "" "{19,19,25,26,16,16,23,28,28,3220,3220,25,3220}" "{o,o,o,o,o,o,o,o,o,o,o,o,o}" "{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn,wal_status,min_keep_lsn}" _null_ _null_ pg_get_replication_slots _null_ _null_ _null_ )); DESCR("information about replication slots currently in use"); DATA(insert OID = 3786 ( pg_create_logical_replication_slot PGNSP PGUID 12 1 0 0 0 f f f f t f v u 3 0 2249 "19 19 16" "{19,19,16,25,3220}" "{i,i,i,o,o}" "{slot_name,plugin,temporary,slot_name,lsn}" _null_ _null_ pg_create_logical_replication_slot _null_ _null_ _null_ )); DESCR("set up a logical replication slot"); diff --git a/src/test/regress/expected/rules.out b/src/test/regress/expected/rules.out index 5433944..c5240c9 100644 --- a/src/test/regress/expected/rules.out +++ b/src/test/regress/expected/rules.out @@ -1451,8 +1451,10 @@ pg_replication_slots| SELECT l.slot_name, l.xmin, l.catalog_xmin, l.restart_lsn, - l.confirmed_flush_lsn - FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn) + l.confirmed_flush_lsn, + l.wal_status, + l.min_keep_lsn + FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn, wal_status, min_keep_lsn) LEFT JOIN pg_database d ON ((l.datoid = d.oid))); pg_roles| SELECT pg_authid.rolname, pg_authid.rolsuper, -- 2.9.2 ----Next_Part(Mon_Jan_29_19_26_34_2018_950)-- Content-Type: Text/X-Patch; charset=us-ascii Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="0003-TAP-test-for-the-slot-limit-feature.patch" ^ permalink raw reply [nested|flat] 13+ messages in thread
* [PATCH 2/4] Add monitoring aid for max_replication_slots. @ 2017-12-21 12:23 Kyotaro Horiguchi <[email protected]> 0 siblings, 0 replies; 13+ messages in thread From: Kyotaro Horiguchi @ 2017-12-21 12:23 UTC (permalink / raw) Adds two columns "status" and "min_secure_lsn" in pg_replication_slot. Setting max_slot_wal_keep_size, long-disconnected slots may lose sync. The two columns shows that a slot can be reconnected or not, or about to lose required WAL segments. And the LSN back to where the next checkpoint will secure. --- contrib/test_decoding/expected/ddl.out | 4 +- src/backend/access/transam/xlog.c | 89 ++++++++++++++++++++++++++++++++++ src/backend/catalog/system_views.sql | 4 +- src/backend/replication/slotfuncs.c | 25 +++++++++- src/include/access/xlog.h | 1 + src/include/catalog/pg_proc.dat | 6 +-- src/test/regress/expected/rules.out | 6 ++- 7 files changed, 126 insertions(+), 9 deletions(-) diff --git a/contrib/test_decoding/expected/ddl.out b/contrib/test_decoding/expected/ddl.out index b7c76469fc..276f7f6efd 100644 --- a/contrib/test_decoding/expected/ddl.out +++ b/contrib/test_decoding/expected/ddl.out @@ -706,7 +706,7 @@ SELECT pg_drop_replication_slot('regression_slot'); /* check that the slot is gone */ SELECT * FROM pg_replication_slots; - slot_name | plugin | slot_type | datoid | database | temporary | active | active_pid | xmin | catalog_xmin | restart_lsn | confirmed_flush_lsn ------------+--------+-----------+--------+----------+-----------+--------+------------+------+--------------+-------------+--------------------- + slot_name | plugin | slot_type | datoid | database | temporary | active | active_pid | xmin | catalog_xmin | restart_lsn | confirmed_flush_lsn | wal_status | min_keep_lsn +-----------+--------+-----------+--------+----------+-----------+--------+------------+------+--------------+-------------+---------------------+------------+-------------- (0 rows) diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index c27a589e89..577be4ecd9 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9426,6 +9426,95 @@ CreateRestartPoint(int flags) return true; } + +/* + * Returns the segment number of the oldest file in XLOG directory. + */ +static XLogSegNo +GetOldestXLogFileSegNo(void) +{ + DIR *xldir; + struct dirent *xlde; + XLogSegNo segno = 0; + + xldir = AllocateDir(XLOGDIR); + while ((xlde = ReadDir(xldir, XLOGDIR)) != NULL) + { + TimeLineID tli; + XLogSegNo fsegno; + + /* Ignore files that are not XLOG segments */ + if (!IsXLogFileName(xlde->d_name) && + !IsPartialXLogFileName(xlde->d_name)) + continue; + + XLogFromFileName(xlde->d_name, &tli, &fsegno, wal_segment_size); + + /* get minimum segment ignorig timeline ID */ + if (segno == 0 || fsegno < segno) + segno = fsegno; + } + + FreeDir(xldir); + + return segno; +} + +/* + * Check if the record on the given restartLSN is present in XLOG files. + * + * Returns true if it is present. If minKeepLSN is given, it receives the + * LSN at the beginning of the oldest existing WAL segment. + */ +bool +IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minKeepLSN) +{ + XLogRecPtr currpos; + XLogSegNo restartSeg; + XLogSegNo tailSeg; + XLogSegNo oldestSeg; + + Assert(!XLogRecPtrIsInvalid(restartLSN)); + + currpos = GetXLogWriteRecPtr(); + + SpinLockAcquire(&XLogCtl->info_lck); + oldestSeg = XLogCtl->lastRemovedSegNo; + SpinLockRelease(&XLogCtl->info_lck); + + /* + * oldestSeg is zero before at least one segment has been removed since + * startup. Use oldest segno taken from file names. + */ + if (oldestSeg == 0) + { + static XLogSegNo oldestFileSeg = 0; + + if (oldestFileSeg == 0) + oldestFileSeg = GetOldestXLogFileSegNo(); + /* let it have the same meaning with lastRemovedSegNo here */ + oldestSeg = oldestFileSeg - 1; + } + + /* oldest segment is just after the last removed segment */ + oldestSeg++; + + XLByteToSeg(restartLSN, restartSeg, wal_segment_size); + + + if (minKeepLSN) + { + XLogRecPtr slotPtr = XLogGetReplicationSlotMinimumLSN(); + Assert(!XLogRecPtrIsInvalid(slotPtr)); + + tailSeg = GetOldestKeepSegment(currpos, slotPtr); + + XLogSegNoOffsetToRecPtr(tailSeg, 0, *minKeepLSN, wal_segment_size); + } + + return oldestSeg <= restartSeg; +} + /* * Returns minimum segment number the next checktpoint must leave considering * wal_keep_segments, replication slots and max_slot_wal_keep_size. diff --git a/src/backend/catalog/system_views.sql b/src/backend/catalog/system_views.sql index 8cd8bf40ac..1664a086e9 100644 --- a/src/backend/catalog/system_views.sql +++ b/src/backend/catalog/system_views.sql @@ -797,7 +797,9 @@ CREATE VIEW pg_replication_slots AS L.xmin, L.catalog_xmin, L.restart_lsn, - L.confirmed_flush_lsn + L.confirmed_flush_lsn, + L.wal_status, + L.min_keep_lsn FROM pg_get_replication_slots() AS L LEFT JOIN pg_database D ON (L.datoid = D.oid); diff --git a/src/backend/replication/slotfuncs.c b/src/backend/replication/slotfuncs.c index 2806e1076c..f13aa4d455 100644 --- a/src/backend/replication/slotfuncs.c +++ b/src/backend/replication/slotfuncs.c @@ -185,7 +185,7 @@ pg_drop_replication_slot(PG_FUNCTION_ARGS) Datum pg_get_replication_slots(PG_FUNCTION_ARGS) { -#define PG_GET_REPLICATION_SLOTS_COLS 11 +#define PG_GET_REPLICATION_SLOTS_COLS 13 ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; TupleDesc tupdesc; Tuplestorestate *tupstore; @@ -307,6 +307,29 @@ pg_get_replication_slots(PG_FUNCTION_ARGS) else nulls[i++] = true; + if (restart_lsn == InvalidXLogRecPtr) + { + values[i++] = CStringGetTextDatum("unknown"); + values[i++] = LSNGetDatum(InvalidXLogRecPtr); + } + else + { + XLogRecPtr min_keep_lsn; + char *status = "lost"; + + if (BoolGetDatum(IsLsnStillAvaiable(restart_lsn, + &min_keep_lsn))) + { + if (min_keep_lsn <= restart_lsn) + status = "streaming"; + else + status = "keeping"; + } + + values[i++] = CStringGetTextDatum(status); + values[i++] = LSNGetDatum(min_keep_lsn); + } + tuplestore_putvalues(tupstore, tupdesc, values, nulls); } LWLockRelease(ReplicationSlotControlLock); diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h index 12cd0d1d10..52e64f392d 100644 --- a/src/include/access/xlog.h +++ b/src/include/access/xlog.h @@ -269,6 +269,7 @@ extern void ShutdownXLOG(int code, Datum arg); extern void InitXLOGAccess(void); extern void CreateCheckPoint(int flags); extern bool CreateRestartPoint(int flags); +extern bool IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minSecureLSN); extern void XLogPutNextOid(Oid nextOid); extern XLogRecPtr XLogRestorePoint(const char *rpName); extern void UpdateFullPageWrites(void); diff --git a/src/include/catalog/pg_proc.dat b/src/include/catalog/pg_proc.dat index 40d54ed030..ef8f0eab91 100644 --- a/src/include/catalog/pg_proc.dat +++ b/src/include/catalog/pg_proc.dat @@ -9796,9 +9796,9 @@ proname => 'pg_get_replication_slots', prorows => '10', proisstrict => 'f', proretset => 't', provolatile => 's', prorettype => 'record', proargtypes => '', - proallargtypes => '{name,name,text,oid,bool,bool,int4,xid,xid,pg_lsn,pg_lsn}', - proargmodes => '{o,o,o,o,o,o,o,o,o,o,o}', - proargnames => '{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn}', + proallargtypes => '{name,name,text,oid,bool,bool,int4,xid,xid,pg_lsn,pg_lsn,text,pg_lsn}', + proargmodes => '{o,o,o,o,o,o,o,o,o,o,o,o,o}', + proargnames => '{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn,wal_status,min_keep_lsn}', prosrc => 'pg_get_replication_slots' }, { oid => '3786', descr => 'set up a logical replication slot', proname => 'pg_create_logical_replication_slot', provolatile => 'v', diff --git a/src/test/regress/expected/rules.out b/src/test/regress/expected/rules.out index ae0cd253d5..93f6bff77e 100644 --- a/src/test/regress/expected/rules.out +++ b/src/test/regress/expected/rules.out @@ -1451,8 +1451,10 @@ pg_replication_slots| SELECT l.slot_name, l.xmin, l.catalog_xmin, l.restart_lsn, - l.confirmed_flush_lsn - FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn) + l.confirmed_flush_lsn, + l.wal_status, + l.min_keep_lsn + FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn, wal_status, min_keep_lsn) LEFT JOIN pg_database d ON ((l.datoid = d.oid))); pg_roles| SELECT pg_authid.rolname, pg_authid.rolsuper, -- 2.16.3 ----Next_Part(Mon_Jul_09_14_47_06_2018_788)-- Content-Type: Text/X-Patch; charset=us-ascii Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="v4-0003-TAP-test-for-the-slot-limit-feature.patch" ^ permalink raw reply [nested|flat] 13+ messages in thread
* [PATCH 2/4] Add monitoring aid for max_replication_slots. @ 2017-12-21 12:23 Kyotaro Horiguchi <[email protected]> 0 siblings, 0 replies; 13+ messages in thread From: Kyotaro Horiguchi @ 2017-12-21 12:23 UTC (permalink / raw) Adds two columns "status" and "min_secure_lsn" in pg_replication_slot. Setting max_slot_wal_keep_size, long-disconnected slots may lose sync. The two columns shows that a slot can be reconnected or not, or about to lose required WAL segments. And the LSN back to where the next checkpoint will secure. --- contrib/test_decoding/expected/ddl.out | 4 +- src/backend/access/transam/xlog.c | 93 ++++++++++++++++++++++++++++++++++ src/backend/catalog/system_views.sql | 4 +- src/backend/replication/slotfuncs.c | 25 ++++++++- src/include/access/xlog.h | 1 + src/include/catalog/pg_proc.dat | 6 +-- src/test/regress/expected/rules.out | 6 ++- 7 files changed, 130 insertions(+), 9 deletions(-) diff --git a/contrib/test_decoding/expected/ddl.out b/contrib/test_decoding/expected/ddl.out index b7c76469fc..276f7f6efd 100644 --- a/contrib/test_decoding/expected/ddl.out +++ b/contrib/test_decoding/expected/ddl.out @@ -706,7 +706,7 @@ SELECT pg_drop_replication_slot('regression_slot'); /* check that the slot is gone */ SELECT * FROM pg_replication_slots; - slot_name | plugin | slot_type | datoid | database | temporary | active | active_pid | xmin | catalog_xmin | restart_lsn | confirmed_flush_lsn ------------+--------+-----------+--------+----------+-----------+--------+------------+------+--------------+-------------+--------------------- + slot_name | plugin | slot_type | datoid | database | temporary | active | active_pid | xmin | catalog_xmin | restart_lsn | confirmed_flush_lsn | wal_status | min_keep_lsn +-----------+--------+-----------+--------+----------+-----------+--------+------------+------+--------------+-------------+---------------------+------------+-------------- (0 rows) diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index cf48ed06af..048f55ab77 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9379,6 +9379,99 @@ CreateRestartPoint(int flags) return true; } + +/* + * Returns the segment number of the oldest file in XLOG directory. + */ +static XLogSegNo +GetOldestXLogFileSegNo(void) +{ + DIR *xldir; + struct dirent *xlde; + XLogSegNo segno = 0; + + xldir = AllocateDir(XLOGDIR); + if (xldir == NULL) + ereport(ERROR, + (errcode_for_file_access(), + errmsg("could not open write-ahead log directory \"%s\": %m", + XLOGDIR))); + + while ((xlde = ReadDir(xldir, XLOGDIR)) != NULL) + { + TimeLineID tli; + XLogSegNo fsegno; + + /* Ignore files that are not XLOG segments */ + if (!IsXLogFileName(xlde->d_name) && + !IsPartialXLogFileName(xlde->d_name)) + continue; + + XLogFromFileName(xlde->d_name, &tli, &fsegno, wal_segment_size); + + /* get minimum segment ignorig timeline ID */ + if (segno == 0 || fsegno < segno) + segno = fsegno; + } + + return segno; +} + +/* + * Check if the record on the given restartLSN is present in XLOG files. + * + * Returns true if it is present. If minKeepLSN is given, it receives the + * LSN at the beginning of the oldest existing WAL segment. + */ +bool +IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minKeepLSN) +{ + XLogRecPtr currpos; + XLogSegNo restartSeg; + XLogSegNo tailSeg; + XLogSegNo oldestSeg; + + Assert(!XLogRecPtrIsInvalid(restartLSN)); + + currpos = GetXLogWriteRecPtr(); + + SpinLockAcquire(&XLogCtl->info_lck); + oldestSeg = XLogCtl->lastRemovedSegNo; + SpinLockRelease(&XLogCtl->info_lck); + + /* + * oldestSeg is zero before at least one segment has been removed since + * startup. Use oldest segno taken from file names. + */ + if (oldestSeg == 0) + { + static XLogSegNo oldestFileSeg = 0; + + if (oldestFileSeg == 0) + oldestFileSeg = GetOldestXLogFileSegNo(); + /* let it have the same meaning with lastRemovedSegNo here */ + oldestSeg = oldestFileSeg - 1; + } + + /* oldest segment is just after the last removed segment */ + oldestSeg++; + + XLByteToSeg(restartLSN, restartSeg, wal_segment_size); + + + if (minKeepLSN) + { + XLogRecPtr slotPtr = XLogGetReplicationSlotMinimumLSN(); + Assert(!XLogRecPtrIsInvalid(slotPtr)); + + tailSeg = GetOldestKeepSegment(currpos, slotPtr); + + XLogSegNoOffsetToRecPtr(tailSeg, 0, *minKeepLSN, wal_segment_size); + } + + return oldestSeg <= restartSeg; +} + /* * Returns minimum segment number the next checktpoint must leave considering * wal_keep_segments, replication slots and max_slot_wal_keep_size. diff --git a/src/backend/catalog/system_views.sql b/src/backend/catalog/system_views.sql index 8cd8bf40ac..1664a086e9 100644 --- a/src/backend/catalog/system_views.sql +++ b/src/backend/catalog/system_views.sql @@ -797,7 +797,9 @@ CREATE VIEW pg_replication_slots AS L.xmin, L.catalog_xmin, L.restart_lsn, - L.confirmed_flush_lsn + L.confirmed_flush_lsn, + L.wal_status, + L.min_keep_lsn FROM pg_get_replication_slots() AS L LEFT JOIN pg_database D ON (L.datoid = D.oid); diff --git a/src/backend/replication/slotfuncs.c b/src/backend/replication/slotfuncs.c index 2806e1076c..f13aa4d455 100644 --- a/src/backend/replication/slotfuncs.c +++ b/src/backend/replication/slotfuncs.c @@ -185,7 +185,7 @@ pg_drop_replication_slot(PG_FUNCTION_ARGS) Datum pg_get_replication_slots(PG_FUNCTION_ARGS) { -#define PG_GET_REPLICATION_SLOTS_COLS 11 +#define PG_GET_REPLICATION_SLOTS_COLS 13 ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; TupleDesc tupdesc; Tuplestorestate *tupstore; @@ -307,6 +307,29 @@ pg_get_replication_slots(PG_FUNCTION_ARGS) else nulls[i++] = true; + if (restart_lsn == InvalidXLogRecPtr) + { + values[i++] = CStringGetTextDatum("unknown"); + values[i++] = LSNGetDatum(InvalidXLogRecPtr); + } + else + { + XLogRecPtr min_keep_lsn; + char *status = "lost"; + + if (BoolGetDatum(IsLsnStillAvaiable(restart_lsn, + &min_keep_lsn))) + { + if (min_keep_lsn <= restart_lsn) + status = "streaming"; + else + status = "keeping"; + } + + values[i++] = CStringGetTextDatum(status); + values[i++] = LSNGetDatum(min_keep_lsn); + } + tuplestore_putvalues(tupstore, tupdesc, values, nulls); } LWLockRelease(ReplicationSlotControlLock); diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h index 12cd0d1d10..52e64f392d 100644 --- a/src/include/access/xlog.h +++ b/src/include/access/xlog.h @@ -269,6 +269,7 @@ extern void ShutdownXLOG(int code, Datum arg); extern void InitXLOGAccess(void); extern void CreateCheckPoint(int flags); extern bool CreateRestartPoint(int flags); +extern bool IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minSecureLSN); extern void XLogPutNextOid(Oid nextOid); extern XLogRecPtr XLogRestorePoint(const char *rpName); extern void UpdateFullPageWrites(void); diff --git a/src/include/catalog/pg_proc.dat b/src/include/catalog/pg_proc.dat index 40d54ed030..ef8f0eab91 100644 --- a/src/include/catalog/pg_proc.dat +++ b/src/include/catalog/pg_proc.dat @@ -9796,9 +9796,9 @@ proname => 'pg_get_replication_slots', prorows => '10', proisstrict => 'f', proretset => 't', provolatile => 's', prorettype => 'record', proargtypes => '', - proallargtypes => '{name,name,text,oid,bool,bool,int4,xid,xid,pg_lsn,pg_lsn}', - proargmodes => '{o,o,o,o,o,o,o,o,o,o,o}', - proargnames => '{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn}', + proallargtypes => '{name,name,text,oid,bool,bool,int4,xid,xid,pg_lsn,pg_lsn,text,pg_lsn}', + proargmodes => '{o,o,o,o,o,o,o,o,o,o,o,o,o}', + proargnames => '{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn,wal_status,min_keep_lsn}', prosrc => 'pg_get_replication_slots' }, { oid => '3786', descr => 'set up a logical replication slot', proname => 'pg_create_logical_replication_slot', provolatile => 'v', diff --git a/src/test/regress/expected/rules.out b/src/test/regress/expected/rules.out index ae0cd253d5..93f6bff77e 100644 --- a/src/test/regress/expected/rules.out +++ b/src/test/regress/expected/rules.out @@ -1451,8 +1451,10 @@ pg_replication_slots| SELECT l.slot_name, l.xmin, l.catalog_xmin, l.restart_lsn, - l.confirmed_flush_lsn - FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn) + l.confirmed_flush_lsn, + l.wal_status, + l.min_keep_lsn + FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn, wal_status, min_keep_lsn) LEFT JOIN pg_database d ON ((l.datoid = d.oid))); pg_roles| SELECT pg_authid.rolname, pg_authid.rolsuper, -- 2.16.3 ----Next_Part(Tue_Jun_26_16_26_59_2018_841)-- Content-Type: Text/X-Patch; charset=us-ascii Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="v2-0001-Add-WAL-releaf-vent-for-replication-slots.patch" ^ permalink raw reply [nested|flat] 13+ messages in thread
* [PATCH 2/4] Add monitoring aid for max_replication_slots. @ 2017-12-21 12:23 Kyotaro Horiguchi <[email protected]> 0 siblings, 0 replies; 13+ messages in thread From: Kyotaro Horiguchi @ 2017-12-21 12:23 UTC (permalink / raw) Adds two columns "status" and "min_secure_lsn" in pg_replication_slot. Setting max_slot_wal_keep_size, long-disconnected slots may lose sync. The two columns shows that a slot can be reconnected or not, or about to lose required WAL segments. And the LSN back to where the next checkpoint will secure. --- contrib/test_decoding/expected/ddl.out | 4 +- src/backend/access/transam/xlog.c | 93 ++++++++++++++++++++++++++++++++++ src/backend/catalog/system_views.sql | 4 +- src/backend/replication/slotfuncs.c | 25 ++++++++- src/include/access/xlog.h | 1 + src/include/catalog/pg_proc.h | 2 +- src/test/regress/expected/rules.out | 6 ++- 7 files changed, 128 insertions(+), 7 deletions(-) diff --git a/contrib/test_decoding/expected/ddl.out b/contrib/test_decoding/expected/ddl.out index 1e22c1eefc..92cd56a5f0 100644 --- a/contrib/test_decoding/expected/ddl.out +++ b/contrib/test_decoding/expected/ddl.out @@ -702,7 +702,7 @@ SELECT pg_drop_replication_slot('regression_slot'); /* check that the slot is gone */ SELECT * FROM pg_replication_slots; - slot_name | plugin | slot_type | datoid | database | temporary | active | active_pid | xmin | catalog_xmin | restart_lsn | confirmed_flush_lsn ------------+--------+-----------+--------+----------+-----------+--------+------------+------+--------------+-------------+--------------------- + slot_name | plugin | slot_type | datoid | database | temporary | active | active_pid | xmin | catalog_xmin | restart_lsn | confirmed_flush_lsn | wal_status | min_keep_lsn +-----------+--------+-----------+--------+----------+-----------+--------+------------+------+--------------+-------------+---------------------+------------+-------------- (0 rows) diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 542e1d78fe..529aee9014 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9346,6 +9346,99 @@ CreateRestartPoint(int flags) return true; } + +/* + * Returns the segment number of the oldest file in XLOG directory. + */ +static XLogSegNo +GetOldestXLogFileSegNo(void) +{ + DIR *xldir; + struct dirent *xlde; + XLogSegNo segno = 0; + + xldir = AllocateDir(XLOGDIR); + if (xldir == NULL) + ereport(ERROR, + (errcode_for_file_access(), + errmsg("could not open write-ahead log directory \"%s\": %m", + XLOGDIR))); + + while ((xlde = ReadDir(xldir, XLOGDIR)) != NULL) + { + TimeLineID tli; + XLogSegNo fsegno; + + /* Ignore files that are not XLOG segments */ + if (!IsXLogFileName(xlde->d_name) && + !IsPartialXLogFileName(xlde->d_name)) + continue; + + XLogFromFileName(xlde->d_name, &tli, &fsegno, wal_segment_size); + + /* get minimum segment ignorig timeline ID */ + if (segno == 0 || fsegno < segno) + segno = fsegno; + } + + return segno; +} + +/* + * Check if the record on the given restartLSN is present in XLOG files. + * + * Returns true if it is present. If minKeepLSN is given, it receives the + * LSN at the beginning of the oldest existing WAL segment. + */ +bool +IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minKeepLSN) +{ + XLogRecPtr currpos; + XLogSegNo restartSeg; + XLogSegNo tailSeg; + XLogSegNo oldestSeg; + + Assert(!XLogRecPtrIsInvalid(restartLSN)); + + currpos = GetXLogWriteRecPtr(); + + SpinLockAcquire(&XLogCtl->info_lck); + oldestSeg = XLogCtl->lastRemovedSegNo; + SpinLockRelease(&XLogCtl->info_lck); + + /* + * oldestSeg is zero before at least one segment has been removed since + * startup. Use oldest segno taken from file names. + */ + if (oldestSeg == 0) + { + static XLogSegNo oldestFileSeg = 0; + + if (oldestFileSeg == 0) + oldestFileSeg = GetOldestXLogFileSegNo(); + /* let it have the same meaning with lastRemovedSegNo here */ + oldestSeg = oldestFileSeg - 1; + } + + /* oldest segment is just after the last removed segment */ + oldestSeg++; + + XLByteToSeg(restartLSN, restartSeg, wal_segment_size); + + + if (minKeepLSN) + { + XLogRecPtr slotPtr = XLogGetReplicationSlotMinimumLSN(); + Assert(!XLogRecPtrIsInvalid(slotPtr)); + + tailSeg = GetOldestKeepSegment(currpos, slotPtr); + + XLogSegNoOffsetToRecPtr(tailSeg, 0, *minKeepLSN, wal_segment_size); + } + + return oldestSeg <= restartSeg; +} + /* * Returns minimum segment number the next checktpoint must leave considering * wal_keep_segments, replication slots and max_slot_wal_keep_size. diff --git a/src/backend/catalog/system_views.sql b/src/backend/catalog/system_views.sql index 5e6e8a64f6..9284175f7d 100644 --- a/src/backend/catalog/system_views.sql +++ b/src/backend/catalog/system_views.sql @@ -795,7 +795,9 @@ CREATE VIEW pg_replication_slots AS L.xmin, L.catalog_xmin, L.restart_lsn, - L.confirmed_flush_lsn + L.confirmed_flush_lsn, + L.wal_status, + L.min_keep_lsn FROM pg_get_replication_slots() AS L LEFT JOIN pg_database D ON (L.datoid = D.oid); diff --git a/src/backend/replication/slotfuncs.c b/src/backend/replication/slotfuncs.c index e873dd1f81..16575fa411 100644 --- a/src/backend/replication/slotfuncs.c +++ b/src/backend/replication/slotfuncs.c @@ -185,7 +185,7 @@ pg_drop_replication_slot(PG_FUNCTION_ARGS) Datum pg_get_replication_slots(PG_FUNCTION_ARGS) { -#define PG_GET_REPLICATION_SLOTS_COLS 11 +#define PG_GET_REPLICATION_SLOTS_COLS 13 ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; TupleDesc tupdesc; Tuplestorestate *tupstore; @@ -307,6 +307,29 @@ pg_get_replication_slots(PG_FUNCTION_ARGS) else nulls[i++] = true; + if (restart_lsn == InvalidXLogRecPtr) + { + values[i++] = CStringGetTextDatum("unknown"); + values[i++] = LSNGetDatum(InvalidXLogRecPtr); + } + else + { + XLogRecPtr min_keep_lsn; + char *status = "lost"; + + if (BoolGetDatum(IsLsnStillAvaiable(restart_lsn, + &min_keep_lsn))) + { + if (min_keep_lsn <= restart_lsn) + status = "streaming"; + else + status = "keeping"; + } + + values[i++] = CStringGetTextDatum(status); + values[i++] = LSNGetDatum(min_keep_lsn); + } + tuplestore_putvalues(tupstore, tupdesc, values, nulls); } LWLockRelease(ReplicationSlotControlLock); diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h index 12cd0d1d10..52e64f392d 100644 --- a/src/include/access/xlog.h +++ b/src/include/access/xlog.h @@ -269,6 +269,7 @@ extern void ShutdownXLOG(int code, Datum arg); extern void InitXLOGAccess(void); extern void CreateCheckPoint(int flags); extern bool CreateRestartPoint(int flags); +extern bool IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minSecureLSN); extern void XLogPutNextOid(Oid nextOid); extern XLogRecPtr XLogRestorePoint(const char *rpName); extern void UpdateFullPageWrites(void); diff --git a/src/include/catalog/pg_proc.h b/src/include/catalog/pg_proc.h index 0fdb42f639..e8e32c1a97 100644 --- a/src/include/catalog/pg_proc.h +++ b/src/include/catalog/pg_proc.h @@ -5385,7 +5385,7 @@ DATA(insert OID = 3779 ( pg_create_physical_replication_slot PGNSP PGUID 12 1 0 DESCR("create a physical replication slot"); DATA(insert OID = 3780 ( pg_drop_replication_slot PGNSP PGUID 12 1 0 0 0 f f f t f v u 1 0 2278 "19" _null_ _null_ _null_ _null_ _null_ pg_drop_replication_slot _null_ _null_ _null_ )); DESCR("drop a replication slot"); -DATA(insert OID = 3781 ( pg_get_replication_slots PGNSP PGUID 12 1 10 0 0 f f f f t s s 0 0 2249 "" "{19,19,25,26,16,16,23,28,28,3220,3220}" "{o,o,o,o,o,o,o,o,o,o,o}" "{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn}" _null_ _null_ pg_get_replication_slots _null_ _null_ _null_ )); +DATA(insert OID = 3781 ( pg_get_replication_slots PGNSP PGUID 12 1 10 0 0 f f f f t s s 0 0 2249 "" "{19,19,25,26,16,16,23,28,28,3220,3220,25,3220}" "{o,o,o,o,o,o,o,o,o,o,o,o,o}" "{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn,wal_status,min_keep_lsn}" _null_ _null_ pg_get_replication_slots _null_ _null_ _null_ )); DESCR("information about replication slots currently in use"); DATA(insert OID = 3786 ( pg_create_logical_replication_slot PGNSP PGUID 12 1 0 0 0 f f f t f v u 3 0 2249 "19 19 16" "{19,19,16,25,3220}" "{i,i,i,o,o}" "{slot_name,plugin,temporary,slot_name,lsn}" _null_ _null_ pg_create_logical_replication_slot _null_ _null_ _null_ )); DESCR("set up a logical replication slot"); diff --git a/src/test/regress/expected/rules.out b/src/test/regress/expected/rules.out index 5e0597e091..3944e36681 100644 --- a/src/test/regress/expected/rules.out +++ b/src/test/regress/expected/rules.out @@ -1451,8 +1451,10 @@ pg_replication_slots| SELECT l.slot_name, l.xmin, l.catalog_xmin, l.restart_lsn, - l.confirmed_flush_lsn - FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn) + l.confirmed_flush_lsn, + l.wal_status, + l.min_keep_lsn + FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn, wal_status, min_keep_lsn) LEFT JOIN pg_database d ON ((l.datoid = d.oid))); pg_roles| SELECT pg_authid.rolname, pg_authid.rolsuper, -- 2.16.2 ----Next_Part(Mon_Mar_19_17_09_48_2018_369)-- Content-Type: Text/X-Patch; charset=us-ascii Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="0003-TAP-test-for-the-slot-limit-feature.patch" ^ permalink raw reply [nested|flat] 13+ messages in thread
* [PATCH 2/4] Add monitoring aid for max_replication_slots. @ 2017-12-21 12:23 Kyotaro Horiguchi <[email protected]> 0 siblings, 0 replies; 13+ messages in thread From: Kyotaro Horiguchi @ 2017-12-21 12:23 UTC (permalink / raw) Adds two columns "status" and "min_secure_lsn" in pg_replication_slot. Setting max_slot_wal_keep_size, long-disconnected slots may lose sync. The two columns shows that a slot can be reconnected or not, or about to lose required WAL segments. And the LSN back to where the next checkpoint will secure. --- contrib/test_decoding/expected/ddl.out | 4 +- src/backend/access/transam/xlog.c | 93 ++++++++++++++++++++++++++++++++++ src/backend/catalog/system_views.sql | 4 +- src/backend/replication/slotfuncs.c | 25 ++++++++- src/include/access/xlog.h | 1 + src/include/catalog/pg_proc.h | 2 +- src/test/regress/expected/rules.out | 6 ++- 7 files changed, 128 insertions(+), 7 deletions(-) diff --git a/contrib/test_decoding/expected/ddl.out b/contrib/test_decoding/expected/ddl.out index 1e22c1e..92cd56a 100644 --- a/contrib/test_decoding/expected/ddl.out +++ b/contrib/test_decoding/expected/ddl.out @@ -702,7 +702,7 @@ SELECT pg_drop_replication_slot('regression_slot'); /* check that the slot is gone */ SELECT * FROM pg_replication_slots; - slot_name | plugin | slot_type | datoid | database | temporary | active | active_pid | xmin | catalog_xmin | restart_lsn | confirmed_flush_lsn ------------+--------+-----------+--------+----------+-----------+--------+------------+------+--------------+-------------+--------------------- + slot_name | plugin | slot_type | datoid | database | temporary | active | active_pid | xmin | catalog_xmin | restart_lsn | confirmed_flush_lsn | wal_status | min_keep_lsn +-----------+--------+-----------+--------+----------+-----------+--------+------------+------+--------------+-------------+---------------------+------------+-------------- (0 rows) diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index c01d0b3..099d89f 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9349,6 +9349,99 @@ CreateRestartPoint(int flags) return true; } + +/* + * Returns the segment number of the oldest file in XLOG directory. + */ +static XLogSegNo +GetOldestXLogFileSegNo(void) +{ + DIR *xldir; + struct dirent *xlde; + XLogSegNo segno = 0; + + xldir = AllocateDir(XLOGDIR); + if (xldir == NULL) + ereport(ERROR, + (errcode_for_file_access(), + errmsg("could not open write-ahead log directory \"%s\": %m", + XLOGDIR))); + + while ((xlde = ReadDir(xldir, XLOGDIR)) != NULL) + { + TimeLineID tli; + XLogSegNo fsegno; + + /* Ignore files that are not XLOG segments */ + if (!IsXLogFileName(xlde->d_name) && + !IsPartialXLogFileName(xlde->d_name)) + continue; + + XLogFromFileName(xlde->d_name, &tli, &fsegno, wal_segment_size); + + /* get minimum segment ignorig timeline ID */ + if (segno == 0 || fsegno < segno) + segno = fsegno; + } + + return segno; +} + +/* + * Check if the record on the given restartLSN is present in XLOG files. + * + * Returns true if it is present. If minKeepLSN is given, it receives the + * LSN at the beginning of the oldest existing WAL segment. + */ +bool +IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minKeepLSN) +{ + XLogRecPtr currpos; + XLogSegNo restartSeg; + XLogSegNo tailSeg; + XLogSegNo oldestSeg; + + Assert(!XLogRecPtrIsInvalid(restartLSN)); + + currpos = GetXLogWriteRecPtr(); + + SpinLockAcquire(&XLogCtl->info_lck); + oldestSeg = XLogCtl->lastRemovedSegNo; + SpinLockRelease(&XLogCtl->info_lck); + + /* + * oldestSeg is zero before at least one segment has been removed since + * startup. Use oldest segno taken from file names. + */ + if (oldestSeg == 0) + { + static XLogSegNo oldestFileSeg = 0; + + if (oldestFileSeg == 0) + oldestFileSeg = GetOldestXLogFileSegNo(); + /* let it have the same meaning with lastRemovedSegNo here */ + oldestSeg = oldestFileSeg - 1; + } + + /* oldest segment is just after the last removed segment */ + oldestSeg++; + + XLByteToSeg(restartLSN, restartSeg, wal_segment_size); + + + if (minKeepLSN) + { + XLogRecPtr slotPtr = XLogGetReplicationSlotMinimumLSN(); + Assert(!XLogRecPtrIsInvalid(slotPtr)); + + tailSeg = GetOldestKeepSegment(currpos, slotPtr); + + XLogSegNoOffsetToRecPtr(tailSeg, 0, *minKeepLSN, wal_segment_size); + } + + return oldestSeg <= restartSeg; +} + /* * Returns minimum segment number the next checktpoint must leave considering * wal_keep_segments, replication slots and max_slot_wal_keep_size. diff --git a/src/backend/catalog/system_views.sql b/src/backend/catalog/system_views.sql index 5652e9e..cd714cc 100644 --- a/src/backend/catalog/system_views.sql +++ b/src/backend/catalog/system_views.sql @@ -793,7 +793,9 @@ CREATE VIEW pg_replication_slots AS L.xmin, L.catalog_xmin, L.restart_lsn, - L.confirmed_flush_lsn + L.confirmed_flush_lsn, + L.wal_status, + L.min_keep_lsn FROM pg_get_replication_slots() AS L LEFT JOIN pg_database D ON (L.datoid = D.oid); diff --git a/src/backend/replication/slotfuncs.c b/src/backend/replication/slotfuncs.c index cf2195b..e6e0386 100644 --- a/src/backend/replication/slotfuncs.c +++ b/src/backend/replication/slotfuncs.c @@ -185,7 +185,7 @@ pg_drop_replication_slot(PG_FUNCTION_ARGS) Datum pg_get_replication_slots(PG_FUNCTION_ARGS) { -#define PG_GET_REPLICATION_SLOTS_COLS 11 +#define PG_GET_REPLICATION_SLOTS_COLS 13 ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; TupleDesc tupdesc; Tuplestorestate *tupstore; @@ -307,6 +307,29 @@ pg_get_replication_slots(PG_FUNCTION_ARGS) else nulls[i++] = true; + if (restart_lsn == InvalidXLogRecPtr) + { + values[i++] = CStringGetTextDatum("unknown"); + values[i++] = LSNGetDatum(InvalidXLogRecPtr); + } + else + { + XLogRecPtr min_keep_lsn; + char *status = "lost"; + + if (BoolGetDatum(IsLsnStillAvaiable(restart_lsn, + &min_keep_lsn))) + { + if (min_keep_lsn <= restart_lsn) + status = "streaming"; + else + status = "keeping"; + } + + values[i++] = CStringGetTextDatum(status); + values[i++] = LSNGetDatum(min_keep_lsn); + } + tuplestore_putvalues(tupstore, tupdesc, values, nulls); } LWLockRelease(ReplicationSlotControlLock); diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h index 12cd0d1..52e64f3 100644 --- a/src/include/access/xlog.h +++ b/src/include/access/xlog.h @@ -269,6 +269,7 @@ extern void ShutdownXLOG(int code, Datum arg); extern void InitXLOGAccess(void); extern void CreateCheckPoint(int flags); extern bool CreateRestartPoint(int flags); +extern bool IsLsnStillAvaiable(XLogRecPtr restartLSN, XLogRecPtr *minSecureLSN); extern void XLogPutNextOid(Oid nextOid); extern XLogRecPtr XLogRestorePoint(const char *rpName); extern void UpdateFullPageWrites(void); diff --git a/src/include/catalog/pg_proc.h b/src/include/catalog/pg_proc.h index f01648c..0712e60 100644 --- a/src/include/catalog/pg_proc.h +++ b/src/include/catalog/pg_proc.h @@ -5345,7 +5345,7 @@ DATA(insert OID = 3779 ( pg_create_physical_replication_slot PGNSP PGUID 12 1 0 DESCR("create a physical replication slot"); DATA(insert OID = 3780 ( pg_drop_replication_slot PGNSP PGUID 12 1 0 0 0 f f f f t f v u 1 0 2278 "19" _null_ _null_ _null_ _null_ _null_ pg_drop_replication_slot _null_ _null_ _null_ )); DESCR("drop a replication slot"); -DATA(insert OID = 3781 ( pg_get_replication_slots PGNSP PGUID 12 1 10 0 0 f f f f f t s s 0 0 2249 "" "{19,19,25,26,16,16,23,28,28,3220,3220}" "{o,o,o,o,o,o,o,o,o,o,o}" "{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn}" _null_ _null_ pg_get_replication_slots _null_ _null_ _null_ )); +DATA(insert OID = 3781 ( pg_get_replication_slots PGNSP PGUID 12 1 10 0 0 f f f f f t s s 0 0 2249 "" "{19,19,25,26,16,16,23,28,28,3220,3220,25,3220}" "{o,o,o,o,o,o,o,o,o,o,o,o,o}" "{slot_name,plugin,slot_type,datoid,temporary,active,active_pid,xmin,catalog_xmin,restart_lsn,confirmed_flush_lsn,wal_status,min_keep_lsn}" _null_ _null_ pg_get_replication_slots _null_ _null_ _null_ )); DESCR("information about replication slots currently in use"); DATA(insert OID = 3786 ( pg_create_logical_replication_slot PGNSP PGUID 12 1 0 0 0 f f f f t f v u 3 0 2249 "19 19 16" "{19,19,16,25,3220}" "{i,i,i,o,o}" "{slot_name,plugin,temporary,slot_name,lsn}" _null_ _null_ pg_create_logical_replication_slot _null_ _null_ _null_ )); DESCR("set up a logical replication slot"); diff --git a/src/test/regress/expected/rules.out b/src/test/regress/expected/rules.out index 5433944..c5240c9 100644 --- a/src/test/regress/expected/rules.out +++ b/src/test/regress/expected/rules.out @@ -1451,8 +1451,10 @@ pg_replication_slots| SELECT l.slot_name, l.xmin, l.catalog_xmin, l.restart_lsn, - l.confirmed_flush_lsn - FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn) + l.confirmed_flush_lsn, + l.wal_status, + l.min_keep_lsn + FROM (pg_get_replication_slots() l(slot_name, plugin, slot_type, datoid, temporary, active, active_pid, xmin, catalog_xmin, restart_lsn, confirmed_flush_lsn, wal_status, min_keep_lsn) LEFT JOIN pg_database d ON ((l.datoid = d.oid))); pg_roles| SELECT pg_authid.rolname, pg_authid.rolsuper, -- 2.9.2 ----Next_Part(Mon_Jan_29_19_40_23_2018_721)-- Content-Type: Text/X-Patch; charset=us-ascii Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="0003-TAP-test-for-the-slot-limit-feature.patch" ^ permalink raw reply [nested|flat] 13+ messages in thread
* [PATCH v1 1/2] WIP: Change instr_time to just store nanoseconds, that's cheaper. @ 2020-05-28 21:52 Andres Freund <[email protected]> 0 siblings, 0 replies; 13+ messages in thread From: Andres Freund @ 2020-05-28 21:52 UTC (permalink / raw) Author: Reviewed-By: Discussion: https://postgr.es/m/ Backpatch: --- src/include/portability/instr_time.h | 62 ++++++++++++---------------- 1 file changed, 26 insertions(+), 36 deletions(-) diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h index d6459327ccf..fc058d548a8 100644 --- a/src/include/portability/instr_time.h +++ b/src/include/portability/instr_time.h @@ -83,63 +83,53 @@ #define PG_INSTR_CLOCK CLOCK_REALTIME #endif -typedef struct timespec instr_time; +typedef int64 instr_time; +#define NS_PER_S INT64CONST(1000000000) +#define US_PER_S INT64CONST(1000000) +#define MS_PER_S INT64CONST(1000) -#define INSTR_TIME_IS_ZERO(t) ((t).tv_nsec == 0 && (t).tv_sec == 0) +#define NS_PER_MS INT64CONST(1000000) +#define NS_PER_US INT64CONST(1000) -#define INSTR_TIME_SET_ZERO(t) ((t).tv_sec = 0, (t).tv_nsec = 0) +#define INSTR_TIME_IS_ZERO(t) ((t) == 0) -#define INSTR_TIME_SET_CURRENT(t) ((void) clock_gettime(PG_INSTR_CLOCK, &(t))) +#define INSTR_TIME_SET_ZERO(t) ((t) = 0) + +static inline instr_time pg_clock_gettime_ns(void) +{ + struct timespec tmp; + + clock_gettime(PG_INSTR_CLOCK, &tmp); + + return tmp.tv_sec * NS_PER_S + tmp.tv_nsec; +} + +#define INSTR_TIME_SET_CURRENT(t) \ + (t) = pg_clock_gettime_ns() #define INSTR_TIME_ADD(x,y) \ do { \ - (x).tv_sec += (y).tv_sec; \ - (x).tv_nsec += (y).tv_nsec; \ - /* Normalize */ \ - while ((x).tv_nsec >= 1000000000) \ - { \ - (x).tv_nsec -= 1000000000; \ - (x).tv_sec++; \ - } \ + (x) += (y); \ } while (0) #define INSTR_TIME_SUBTRACT(x,y) \ do { \ - (x).tv_sec -= (y).tv_sec; \ - (x).tv_nsec -= (y).tv_nsec; \ - /* Normalize */ \ - while ((x).tv_nsec < 0) \ - { \ - (x).tv_nsec += 1000000000; \ - (x).tv_sec--; \ - } \ + (x) -= (y); \ } while (0) #define INSTR_TIME_ACCUM_DIFF(x,y,z) \ do { \ - (x).tv_sec += (y).tv_sec - (z).tv_sec; \ - (x).tv_nsec += (y).tv_nsec - (z).tv_nsec; \ - /* Normalize after each add to avoid overflow/underflow of tv_nsec */ \ - while ((x).tv_nsec < 0) \ - { \ - (x).tv_nsec += 1000000000; \ - (x).tv_sec--; \ - } \ - while ((x).tv_nsec >= 1000000000) \ - { \ - (x).tv_nsec -= 1000000000; \ - (x).tv_sec++; \ - } \ + (x) += (y) - (z); \ } while (0) #define INSTR_TIME_GET_DOUBLE(t) \ - (((double) (t).tv_sec) + ((double) (t).tv_nsec) / 1000000000.0) + ((double) (t) / NS_PER_S) #define INSTR_TIME_GET_MILLISEC(t) \ - (((double) (t).tv_sec * 1000.0) + ((double) (t).tv_nsec) / 1000000.0) + ((double) (t) / NS_PER_MS) #define INSTR_TIME_GET_MICROSEC(t) \ - (((uint64) (t).tv_sec * (uint64) 1000000) + (uint64) ((t).tv_nsec / 1000)) + ((double) (t) / NS_PER_US) #else /* !HAVE_CLOCK_GETTIME */ -- 2.25.0.114.g5b0ca878e0 --wn4ncs637ccpvbhb Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v1-0002-WIP-Use-cpu-reference-cycles-via-rdtsc-to-measure.patch" ^ permalink raw reply [nested|flat] 13+ messages in thread
end of thread, other threads:[~2020-05-28 21:52 UTC | newest] Thread overview: 13+ messages (download: mbox mbox.gz follow: Atom feed) -- links below jump to the message on this page -- 2017-09-07 10:13 [PATCH 2/2] Add monitoring aid for max_replication_slots. Kyotaro Horiguchi <[email protected]> 2017-09-07 10:13 [PATCH 2/2] Add monitoring aid for max_replication_slots. Kyotaro Horiguchi <[email protected]> 2017-09-07 10:13 [PATCH 2/2] Add monitoring aid for max_replication_slots. Kyotaro Horiguchi <[email protected]> 2017-09-07 10:13 [PATCH 2/2] Add monitoring aid for max_replication_slots. Kyotaro Horiguchi <[email protected]> 2017-12-21 12:23 [PATCH 2/4] Add monitoring aid for max_replication_slots. Kyotaro Horiguchi <[email protected]> 2017-12-21 12:23 [PATCH 2/4] Add monitoring aid for max_replication_slots. Kyotaro Horiguchi <[email protected]> 2017-12-21 12:23 [PATCH 2/4] Add monitoring aid for max_replication_slots. Kyotaro Horiguchi <[email protected]> 2017-12-21 12:23 [PATCH 2/4] Add monitoring aid for max_replication_slots. Kyotaro Horiguchi <[email protected]> 2017-12-21 12:23 [PATCH 2/4] Add monitoring aid for max_replication_slots. Kyotaro Horiguchi <[email protected]> 2017-12-21 12:23 [PATCH 2/4] Add monitoring aid for max_replication_slots. Kyotaro Horiguchi <[email protected]> 2017-12-21 12:23 [PATCH 2/4] Add monitoring aid for max_replication_slots. Kyotaro Horiguchi <[email protected]> 2017-12-21 12:23 [PATCH 2/3] Add monitoring aid for max_replication_slots. Kyotaro Horiguchi <[email protected]> 2020-05-28 21:52 [PATCH v1 1/2] WIP: Change instr_time to just store nanoseconds, that's cheaper. 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