agora inbox for [email protected]  

help / color / mirror / Atom feed

[PATCH v9 1/2] Key PGSTAT_KIND_RELATION by relfile locator
291+ messages / 2 participants
[nested] [flat]

* [PATCH v15 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

This is possible as, since b14e9ce7d55c, the objoid is now uint64 and spcOid
and relNumber are 32 bits.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/postmaster/autovacuum.c          |  18 +-
 src/backend/utils/activity/pgstat_relation.c | 236 ++++++++++++++++---
 src/backend/utils/adt/pgstatfuncs.c          |  21 +-
 src/include/catalog/pg_tablespace.dat        |   4 +
 src/include/catalog/pg_tablespace.h          |   8 +
 src/include/pgstat.h                         |  15 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 8 files changed, 272 insertions(+), 71 deletions(-)
   6.2% src/backend/postmaster/
  61.7% src/backend/utils/activity/
   4.9% src/backend/utils/adt/
   3.2% src/include/catalog/
   5.6% src/include/
  18.1% src/test/recovery/t/

diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index d695f1de4bd..0a028b18d66 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -2035,6 +2035,7 @@ do_autovacuum(void)
 		bool		doanalyze;
 		bool		wraparound;
 		AutoVacuumScores scores;
+		RelFileLocator locator;
 
 		if (classForm->relkind != RELKIND_RELATION &&
 			classForm->relkind != RELKIND_MATVIEW)
@@ -2042,6 +2043,10 @@ do_autovacuum(void)
 
 		relid = classForm->oid;
 
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
+
 		/*
 		 * Check if it is a temp table (presumably, of some other backend's).
 		 * We cannot safely process other backends' temp tables.
@@ -2069,8 +2074,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2143,6 +2147,7 @@ do_autovacuum(void)
 		bool		doanalyze;
 		bool		wraparound;
 		AutoVacuumScores scores;
+		RelFileLocator locator;
 
 		/*
 		 * We cannot safely process other backends' temp tables, so skip 'em.
@@ -2151,6 +2156,9 @@ do_autovacuum(void)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * fetch reloptions -- if this toast table does not have them, try the
@@ -2170,8 +2178,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2988,8 +2995,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	AutoVacuumScores scores;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index bc8c43b96aa..89bf0cbed56 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,37 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +197,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +215,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -213,20 +251,23 @@ pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
 
+	locator = rel->rd_locator;
+
 	/* Store the data in the table's hash table entry. */
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel), false);
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
+											false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -285,9 +326,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -325,9 +366,25 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -468,7 +525,16 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
+}
+
+PgStat_StatTabEntry *
+pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator)
+{
+	return (PgStat_StatTabEntry *) pgstat_fetch_entry(
+													  PGSTAT_KIND_RELATION,
+													  locator.dbOid,
+													  RelFileLocatorToPgStatObjid(locator));
 }
 
 /*
@@ -476,12 +542,14 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
+
+	return pgstat_fetch_stat_tabentry_by_locator(locator);
 }
 
 /*
@@ -503,14 +571,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -706,8 +777,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -750,7 +825,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -785,8 +860,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -920,17 +995,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1009,3 +1088,82 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index 9185a8e6b83..db882339f1e 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -30,6 +30,7 @@
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 #include "utils/tuplestore.h"
 #include "utils/wait_event.h"
@@ -2004,9 +2005,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2363,5 +2369,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/catalog/pg_tablespace.dat b/src/include/catalog/pg_tablespace.dat
index c4cde415219..73ed046be31 100644
--- a/src/include/catalog/pg_tablespace.dat
+++ b/src/include/catalog/pg_tablespace.dat
@@ -10,6 +10,10 @@
 #
 #----------------------------------------------------------------------
 
+/*
+ * When adding a new one, ensure it does not conflict with
+ * PSEUDO_PARTITION_TABLE_SPCOID.
+ */
 [
 
 { oid => '1663', oid_symbol => 'DEFAULTTABLESPACE_OID',
diff --git a/src/include/catalog/pg_tablespace.h b/src/include/catalog/pg_tablespace.h
index 3bd4a74f003..e5e818eb0c5 100644
--- a/src/include/catalog/pg_tablespace.h
+++ b/src/include/catalog/pg_tablespace.h
@@ -21,6 +21,14 @@
 #include "catalog/genbki.h"
 #include "catalog/pg_tablespace_d.h"	/* IWYU pragma: export */
 
+/*
+ * Reserved tablespace OID for partitioned table pseudo locators.
+ * This is not an actual tablespace, just a reserved value to distinguish
+ * partitioned table statistics from regular table statistics. Ensures it does
+ * not conflict with the ones in pg_tablespace.dat.
+ */
+#define PSEUDO_PARTITION_TABLE_SPCOID 1665
+
 /* ----------------
  *		pg_tablespace definition.  cpp turns this into
  *		typedef struct FormData_pg_tablespace
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index 8e3549c3752..ef740aff26e 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -16,6 +16,7 @@
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
 #include "storage/locktag.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -39,6 +40,12 @@ typedef struct RelationData *Relation;
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -179,11 +186,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -766,8 +773,8 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index eed4c6b359c..1090f049058 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -786,6 +786,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index cdc427dbc78..4d00087dc6f 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--Og8T0Lf2JVsuKOzn
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v15-0002-handle-relation-statistics-correctly-during-rewr.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread

* [PATCH v10 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

This is possible as, since b14e9ce7d55c, the objoid is now uint64 and spcOid
and relNumber are 32 bits.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/postmaster/autovacuum.c          |  17 +-
 src/backend/utils/activity/pgstat_relation.c | 236 ++++++++++++++++---
 src/backend/utils/adt/pgstatfuncs.c          |  22 +-
 src/include/catalog/pg_tablespace.dat        |   4 +
 src/include/catalog/pg_tablespace.h          |   8 +
 src/include/pgstat.h                         |  15 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 8 files changed, 271 insertions(+), 72 deletions(-)
   6.1% src/backend/postmaster/
  61.6% src/backend/utils/activity/
   5.1% src/backend/utils/adt/
   3.2% src/include/catalog/
   5.6% src/include/
  18.1% src/test/recovery/t/

diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index 1bd3924e35e..a11174b25ad 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -2014,12 +2014,16 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		if (classForm->relkind != RELKIND_RELATION &&
 			classForm->relkind != RELKIND_MATVIEW)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * Check if it is a temp table (presumably, of some other backend's).
@@ -2048,8 +2052,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2114,6 +2117,7 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		/*
 		 * We cannot safely process other backends' temp tables, so skip 'em.
@@ -2122,6 +2126,9 @@ do_autovacuum(void)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * fetch reloptions -- if this toast table does not have them, try the
@@ -2141,8 +2148,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2939,8 +2945,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	PgStat_StatTabEntry *tabentry;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index 55a10c299db..a267e93f8be 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,37 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +197,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +215,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -213,20 +251,23 @@ pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
 
+	locator = rel->rd_locator;
+
 	/* Store the data in the table's hash table entry. */
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel), false);
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
+											false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -285,9 +326,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -325,9 +366,25 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -468,7 +525,16 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
+}
+
+PgStat_StatTabEntry *
+pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator)
+{
+	return (PgStat_StatTabEntry *) pgstat_fetch_entry(
+													  PGSTAT_KIND_RELATION,
+													  locator.dbOid,
+													  RelFileLocatorToPgStatObjid(locator));
 }
 
 /*
@@ -476,12 +542,14 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
+
+	return pgstat_fetch_stat_tabentry_by_locator(locator);
 }
 
 /*
@@ -503,14 +571,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -706,8 +777,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -750,7 +825,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -785,8 +860,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -920,17 +995,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1009,3 +1088,82 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index ef6fffe60b9..60ffb1679ec 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -23,13 +23,13 @@
 #include "common/ip.h"
 #include "funcapi.h"
 #include "miscadmin.h"
-#include "pgstat.h"
 #include "postmaster/bgworker.h"
 #include "replication/logicallauncher.h"
 #include "storage/proc.h"
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 
 #define UINT32_ACCESS_ONCE(var)		 ((uint32)(*((volatile uint32 *)&(var))))
@@ -1949,9 +1949,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2305,5 +2310,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/catalog/pg_tablespace.dat b/src/include/catalog/pg_tablespace.dat
index 1302a3d75cd..9430970fffd 100644
--- a/src/include/catalog/pg_tablespace.dat
+++ b/src/include/catalog/pg_tablespace.dat
@@ -10,6 +10,10 @@
 #
 #----------------------------------------------------------------------
 
+/*
+ * When adding a new one, ensure it does not conflict with
+ * PSEUDO_PARTITION_TABLE_SPCOID.
+ */
 [
 
 { oid => '1663', oid_symbol => 'DEFAULTTABLESPACE_OID',
diff --git a/src/include/catalog/pg_tablespace.h b/src/include/catalog/pg_tablespace.h
index 7816d779d8c..0e2d8051d69 100644
--- a/src/include/catalog/pg_tablespace.h
+++ b/src/include/catalog/pg_tablespace.h
@@ -21,6 +21,14 @@
 #include "catalog/genbki.h"
 #include "catalog/pg_tablespace_d.h"	/* IWYU pragma: export */
 
+/*
+ * Reserved tablespace OID for partitioned table pseudo locators.
+ * This is not an actual tablespace, just a reserved value to distinguish
+ * partitioned table statistics from regular table statistics. Ensures it does
+ * not conflict with the ones in pg_tablespace.dat.
+ */
+#define PSEUDO_PARTITION_TABLE_SPCOID 1665
+
 /* ----------------
  *		pg_tablespace definition.  cpp turns this into
  *		typedef struct FormData_pg_tablespace
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index 6714363144a..3102f86aa24 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -17,6 +17,7 @@
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
 #include "replication/worker_internal.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -35,6 +36,12 @@
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -175,11 +182,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -735,8 +742,8 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index 5c1ce4d3d6a..7b24928b00d 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -764,6 +764,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index 021e2bf361f..3a9c05eaf10 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--xJsrX25x3tm7Tn81
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v10-0002-handle-relation-statistics-correctly-during-rewr.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread

* [PATCH v13 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

This is possible as, since b14e9ce7d55c, the objoid is now uint64 and spcOid
and relNumber are 32 bits.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/postmaster/autovacuum.c          |  17 +-
 src/backend/utils/activity/pgstat_relation.c | 236 ++++++++++++++++---
 src/backend/utils/adt/pgstatfuncs.c          |  21 +-
 src/include/catalog/pg_tablespace.dat        |   4 +
 src/include/catalog/pg_tablespace.h          |   8 +
 src/include/pgstat.h                         |  15 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 8 files changed, 271 insertions(+), 71 deletions(-)
   6.1% src/backend/postmaster/
  61.7% src/backend/utils/activity/
   4.9% src/backend/utils/adt/
   3.2% src/include/catalog/
   5.6% src/include/
  18.1% src/test/recovery/t/

diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index 219673db930..4bef7ab7543 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -1992,12 +1992,16 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		if (classForm->relkind != RELKIND_RELATION &&
 			classForm->relkind != RELKIND_MATVIEW)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * Check if it is a temp table (presumably, of some other backend's).
@@ -2026,8 +2030,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2092,6 +2095,7 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		/*
 		 * We cannot safely process other backends' temp tables, so skip 'em.
@@ -2100,6 +2104,9 @@ do_autovacuum(void)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * fetch reloptions -- if this toast table does not have them, try the
@@ -2119,8 +2126,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2917,8 +2923,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	PgStat_StatTabEntry *tabentry;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index bc8c43b96aa..89bf0cbed56 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,37 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +197,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +215,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -213,20 +251,23 @@ pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
 
+	locator = rel->rd_locator;
+
 	/* Store the data in the table's hash table entry. */
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel), false);
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
+											false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -285,9 +326,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -325,9 +366,25 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -468,7 +525,16 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
+}
+
+PgStat_StatTabEntry *
+pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator)
+{
+	return (PgStat_StatTabEntry *) pgstat_fetch_entry(
+													  PGSTAT_KIND_RELATION,
+													  locator.dbOid,
+													  RelFileLocatorToPgStatObjid(locator));
 }
 
 /*
@@ -476,12 +542,14 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
+
+	return pgstat_fetch_stat_tabentry_by_locator(locator);
 }
 
 /*
@@ -503,14 +571,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -706,8 +777,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -750,7 +825,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -785,8 +860,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -920,17 +995,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1009,3 +1088,82 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index 5f907335990..106d1e577a7 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -30,6 +30,7 @@
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 #include "utils/tuplestore.h"
 #include "utils/wait_event.h"
@@ -1965,9 +1966,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2324,5 +2330,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/catalog/pg_tablespace.dat b/src/include/catalog/pg_tablespace.dat
index c4cde415219..73ed046be31 100644
--- a/src/include/catalog/pg_tablespace.dat
+++ b/src/include/catalog/pg_tablespace.dat
@@ -10,6 +10,10 @@
 #
 #----------------------------------------------------------------------
 
+/*
+ * When adding a new one, ensure it does not conflict with
+ * PSEUDO_PARTITION_TABLE_SPCOID.
+ */
 [
 
 { oid => '1663', oid_symbol => 'DEFAULTTABLESPACE_OID',
diff --git a/src/include/catalog/pg_tablespace.h b/src/include/catalog/pg_tablespace.h
index 3bd4a74f003..e5e818eb0c5 100644
--- a/src/include/catalog/pg_tablespace.h
+++ b/src/include/catalog/pg_tablespace.h
@@ -21,6 +21,14 @@
 #include "catalog/genbki.h"
 #include "catalog/pg_tablespace_d.h"	/* IWYU pragma: export */
 
+/*
+ * Reserved tablespace OID for partitioned table pseudo locators.
+ * This is not an actual tablespace, just a reserved value to distinguish
+ * partitioned table statistics from regular table statistics. Ensures it does
+ * not conflict with the ones in pg_tablespace.dat.
+ */
+#define PSEUDO_PARTITION_TABLE_SPCOID 1665
+
 /* ----------------
  *		pg_tablespace definition.  cpp turns this into
  *		typedef struct FormData_pg_tablespace
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index 216b93492ba..689c624f373 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -15,6 +15,7 @@
 #include "portability/instr_time.h"
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -38,6 +39,12 @@ typedef struct RelationData *Relation;
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -178,11 +185,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -738,8 +745,8 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index 9b8fbae00ed..42f29496551 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -765,6 +765,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index cdc427dbc78..4d00087dc6f 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--USlF2Vr25us5UpvY
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v13-0002-handle-relation-statistics-correctly-during-rewr.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread

* [PATCH v13 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

This is possible as, since b14e9ce7d55c, the objoid is now uint64 and spcOid
and relNumber are 32 bits.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/postmaster/autovacuum.c          |  17 +-
 src/backend/utils/activity/pgstat_relation.c | 236 ++++++++++++++++---
 src/backend/utils/adt/pgstatfuncs.c          |  21 +-
 src/include/catalog/pg_tablespace.dat        |   4 +
 src/include/catalog/pg_tablespace.h          |   8 +
 src/include/pgstat.h                         |  15 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 8 files changed, 271 insertions(+), 71 deletions(-)
   6.1% src/backend/postmaster/
  61.7% src/backend/utils/activity/
   4.9% src/backend/utils/adt/
   3.2% src/include/catalog/
   5.6% src/include/
  18.1% src/test/recovery/t/

diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index 219673db930..4bef7ab7543 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -1992,12 +1992,16 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		if (classForm->relkind != RELKIND_RELATION &&
 			classForm->relkind != RELKIND_MATVIEW)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * Check if it is a temp table (presumably, of some other backend's).
@@ -2026,8 +2030,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2092,6 +2095,7 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		/*
 		 * We cannot safely process other backends' temp tables, so skip 'em.
@@ -2100,6 +2104,9 @@ do_autovacuum(void)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * fetch reloptions -- if this toast table does not have them, try the
@@ -2119,8 +2126,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2917,8 +2923,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	PgStat_StatTabEntry *tabentry;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index bc8c43b96aa..89bf0cbed56 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,37 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +197,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +215,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -213,20 +251,23 @@ pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
 
+	locator = rel->rd_locator;
+
 	/* Store the data in the table's hash table entry. */
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel), false);
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
+											false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -285,9 +326,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -325,9 +366,25 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -468,7 +525,16 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
+}
+
+PgStat_StatTabEntry *
+pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator)
+{
+	return (PgStat_StatTabEntry *) pgstat_fetch_entry(
+													  PGSTAT_KIND_RELATION,
+													  locator.dbOid,
+													  RelFileLocatorToPgStatObjid(locator));
 }
 
 /*
@@ -476,12 +542,14 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
+
+	return pgstat_fetch_stat_tabentry_by_locator(locator);
 }
 
 /*
@@ -503,14 +571,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -706,8 +777,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -750,7 +825,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -785,8 +860,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -920,17 +995,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1009,3 +1088,82 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index 5f907335990..106d1e577a7 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -30,6 +30,7 @@
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 #include "utils/tuplestore.h"
 #include "utils/wait_event.h"
@@ -1965,9 +1966,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2324,5 +2330,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/catalog/pg_tablespace.dat b/src/include/catalog/pg_tablespace.dat
index c4cde415219..73ed046be31 100644
--- a/src/include/catalog/pg_tablespace.dat
+++ b/src/include/catalog/pg_tablespace.dat
@@ -10,6 +10,10 @@
 #
 #----------------------------------------------------------------------
 
+/*
+ * When adding a new one, ensure it does not conflict with
+ * PSEUDO_PARTITION_TABLE_SPCOID.
+ */
 [
 
 { oid => '1663', oid_symbol => 'DEFAULTTABLESPACE_OID',
diff --git a/src/include/catalog/pg_tablespace.h b/src/include/catalog/pg_tablespace.h
index 3bd4a74f003..e5e818eb0c5 100644
--- a/src/include/catalog/pg_tablespace.h
+++ b/src/include/catalog/pg_tablespace.h
@@ -21,6 +21,14 @@
 #include "catalog/genbki.h"
 #include "catalog/pg_tablespace_d.h"	/* IWYU pragma: export */
 
+/*
+ * Reserved tablespace OID for partitioned table pseudo locators.
+ * This is not an actual tablespace, just a reserved value to distinguish
+ * partitioned table statistics from regular table statistics. Ensures it does
+ * not conflict with the ones in pg_tablespace.dat.
+ */
+#define PSEUDO_PARTITION_TABLE_SPCOID 1665
+
 /* ----------------
  *		pg_tablespace definition.  cpp turns this into
  *		typedef struct FormData_pg_tablespace
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index 216b93492ba..689c624f373 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -15,6 +15,7 @@
 #include "portability/instr_time.h"
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -38,6 +39,12 @@ typedef struct RelationData *Relation;
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -178,11 +185,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -738,8 +745,8 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index 9b8fbae00ed..42f29496551 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -765,6 +765,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index cdc427dbc78..4d00087dc6f 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--USlF2Vr25us5UpvY
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v13-0002-handle-relation-statistics-correctly-during-rewr.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread

* [PATCH v8 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/access/heap/vacuumlazy.c         |   3 +-
 src/backend/postmaster/autovacuum.c          |   9 +-
 src/backend/utils/activity/pgstat_relation.c | 234 +++++++++++++++----
 src/backend/utils/adt/pgstatfuncs.c          |  22 +-
 src/include/pgstat.h                         |  18 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 7 files changed, 249 insertions(+), 78 deletions(-)
   3.3% src/backend/postmaster/
  64.6% src/backend/utils/activity/
   5.3% src/backend/utils/adt/
   7.0% src/include/
  18.7% src/test/recovery/t/

diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c
index 62035b7f9c3..a9b2b4e1033 100644
--- a/src/backend/access/heap/vacuumlazy.c
+++ b/src/backend/access/heap/vacuumlazy.c
@@ -961,8 +961,7 @@ heap_vacuum_rel(Relation rel, const VacuumParams params,
 	 * soon in cases where the failsafe prevented significant amounts of heap
 	 * vacuuming.
 	 */
-	pgstat_report_vacuum(RelationGetRelid(rel),
-						 rel->rd_rel->relisshared,
+	pgstat_report_vacuum(rel->rd_locator,
 						 Max(vacrel->new_live_tuples, 0),
 						 vacrel->recently_dead_tuples +
 						 vacrel->missed_dead_tuples,
diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index 1bd3924e35e..563a3697690 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -2048,8 +2048,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2141,8 +2140,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2939,8 +2937,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	PgStat_StatTabEntry *tabentry;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index b90754f8578..ee6f2eb2bdb 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,38 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use the relation OID as relNumber.
+		 * No collision with regular relations is possible because relNumbers
+		 * are also assigned from the pg_class OID space (see
+		 * GetNewRelFileNumber()), making each value unique across the
+		 * database regardless of spcOid.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = InvalidOid;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +198,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +216,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -207,14 +246,12 @@ pgstat_drop_relation(Relation rel)
  * Report that the table was just vacuumed and flush IO statistics.
  */
 void
-pgstat_report_vacuum(Oid tableoid, bool shared,
-					 PgStat_Counter livetuples, PgStat_Counter deadtuples,
-					 TimestampTz starttime)
+pgstat_report_vacuum(RelFileLocator locator, PgStat_Counter livetuples,
+					 PgStat_Counter deadtuples, TimestampTz starttime)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
 
@@ -227,7 +264,7 @@ pgstat_report_vacuum(Oid tableoid, bool shared,
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
 	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
-											dboid, tableoid, false);
+											locator.dbOid, RelFileLocatorToPgStatObjid(locator), false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -286,9 +323,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -326,9 +363,26 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use the relation OID as relNumber.
+		 * No collision with regular relations is possible because relNumbers
+		 * are also assigned from the pg_class OID space (see
+		 * GetNewRelFileNumber()), making each value unique across the
+		 * database regardless of spcOid.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = InvalidOid;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -469,7 +523,7 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
 }
 
 /*
@@ -477,12 +531,19 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	PgStat_StatTabEntry *tabentry;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	/* fetch the stats entry using the relfilenode based key */
+	tabentry = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION,
+														  locator.dbOid,
+														  RelFileLocatorToPgStatObjid(locator));
+	return tabentry;
 }
 
 /*
@@ -504,14 +565,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -707,8 +771,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -751,7 +819,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -786,8 +854,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -921,17 +989,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1010,3 +1082,83 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use the relation OID as relNumber.
+		 * No collision with regular relations is possible because relNumbers
+		 * are also assigned from the pg_class OID space (see
+		 * GetNewRelFileNumber()), making each value unique across the
+		 * database regardless of spcOid.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = InvalidOid;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index ef6fffe60b9..60ffb1679ec 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -23,13 +23,13 @@
 #include "common/ip.h"
 #include "funcapi.h"
 #include "miscadmin.h"
-#include "pgstat.h"
 #include "postmaster/bgworker.h"
 #include "replication/logicallauncher.h"
 #include "storage/proc.h"
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 
 #define UINT32_ACCESS_ONCE(var)		 ((uint32)(*((volatile uint32 *)&(var))))
@@ -1949,9 +1949,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2305,5 +2310,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index f23dd5870da..c9e451094f3 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -17,6 +17,7 @@
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
 #include "replication/worker_internal.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -35,6 +36,12 @@
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -175,11 +182,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -669,8 +676,8 @@ extern void pgstat_init_relation(Relation rel);
 extern void pgstat_assoc_relation(Relation rel);
 extern void pgstat_unlink_relation(Relation rel);
 
-extern void pgstat_report_vacuum(Oid tableoid, bool shared,
-								 PgStat_Counter livetuples, PgStat_Counter deadtuples,
+extern void pgstat_report_vacuum(RelFileLocator locator, PgStat_Counter livetuples,
+								 PgStat_Counter deadtuples,
 								 TimestampTz starttime);
 extern void pgstat_report_analyze(Relation rel,
 								  PgStat_Counter livetuples, PgStat_Counter deadtuples,
@@ -735,8 +742,7 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index 5c1ce4d3d6a..7b24928b00d 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -764,6 +764,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index 021e2bf361f..3a9c05eaf10 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--BHD/2LBqtxMXWADJ
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v8-0002-handle-relation-statistics-correctly-during-rewri.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread

* [PATCH v9 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

This is possible as, since b14e9ce7d55c, the objoid is now uint64 and spcOid
and relNumber are 32 bits.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/access/heap/vacuumlazy.c         |   3 +-
 src/backend/postmaster/autovacuum.c          |  17 +-
 src/backend/utils/activity/pgstat_relation.c | 239 +++++++++++++++----
 src/backend/utils/adt/pgstatfuncs.c          |  22 +-
 src/include/catalog/pg_tablespace.dat        |   4 +
 src/include/catalog/pg_tablespace.h          |   8 +
 src/include/pgstat.h                         |  19 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 9 files changed, 275 insertions(+), 78 deletions(-)
   5.9% src/backend/postmaster/
  60.4% src/backend/utils/activity/
   4.9% src/backend/utils/adt/
   3.0% src/include/catalog/
   7.1% src/include/
  17.5% src/test/recovery/t/

diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c
index 62035b7f9c3..30778a15639 100644
--- a/src/backend/access/heap/vacuumlazy.c
+++ b/src/backend/access/heap/vacuumlazy.c
@@ -961,8 +961,7 @@ heap_vacuum_rel(Relation rel, const VacuumParams params,
 	 * soon in cases where the failsafe prevented significant amounts of heap
 	 * vacuuming.
 	 */
-	pgstat_report_vacuum(RelationGetRelid(rel),
-						 rel->rd_rel->relisshared,
+	pgstat_report_vacuum(rel,
 						 Max(vacrel->new_live_tuples, 0),
 						 vacrel->recently_dead_tuples +
 						 vacrel->missed_dead_tuples,
diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index 1bd3924e35e..a11174b25ad 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -2014,12 +2014,16 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		if (classForm->relkind != RELKIND_RELATION &&
 			classForm->relkind != RELKIND_MATVIEW)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * Check if it is a temp table (presumably, of some other backend's).
@@ -2048,8 +2052,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2114,6 +2117,7 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		/*
 		 * We cannot safely process other backends' temp tables, so skip 'em.
@@ -2122,6 +2126,9 @@ do_autovacuum(void)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * fetch reloptions -- if this toast table does not have them, try the
@@ -2141,8 +2148,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2939,8 +2945,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	PgStat_StatTabEntry *tabentry;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index b90754f8578..48bf93cae6e 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,37 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +197,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +215,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -207,27 +245,29 @@ pgstat_drop_relation(Relation rel)
  * Report that the table was just vacuumed and flush IO statistics.
  */
 void
-pgstat_report_vacuum(Oid tableoid, bool shared,
-					 PgStat_Counter livetuples, PgStat_Counter deadtuples,
-					 TimestampTz starttime)
+pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
+					 PgStat_Counter deadtuples, TimestampTz starttime)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
 
+	locator = rel->rd_locator;
 	/* Store the data in the table's hash table entry. */
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
 	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
-											dboid, tableoid, false);
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
+											false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -286,9 +326,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -326,9 +366,25 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -469,7 +525,16 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
+}
+
+PgStat_StatTabEntry *
+pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator)
+{
+	return (PgStat_StatTabEntry *) pgstat_fetch_entry(
+													  PGSTAT_KIND_RELATION,
+													  locator.dbOid,
+													  RelFileLocatorToPgStatObjid(locator));
 }
 
 /*
@@ -477,12 +542,14 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
+
+	return pgstat_fetch_stat_tabentry_by_locator(locator);
 }
 
 /*
@@ -504,14 +571,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -707,8 +777,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -751,7 +825,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -786,8 +860,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -921,17 +995,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1010,3 +1088,82 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index ef6fffe60b9..60ffb1679ec 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -23,13 +23,13 @@
 #include "common/ip.h"
 #include "funcapi.h"
 #include "miscadmin.h"
-#include "pgstat.h"
 #include "postmaster/bgworker.h"
 #include "replication/logicallauncher.h"
 #include "storage/proc.h"
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 
 #define UINT32_ACCESS_ONCE(var)		 ((uint32)(*((volatile uint32 *)&(var))))
@@ -1949,9 +1949,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2305,5 +2310,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/catalog/pg_tablespace.dat b/src/include/catalog/pg_tablespace.dat
index 1302a3d75cd..9430970fffd 100644
--- a/src/include/catalog/pg_tablespace.dat
+++ b/src/include/catalog/pg_tablespace.dat
@@ -10,6 +10,10 @@
 #
 #----------------------------------------------------------------------
 
+/*
+ * When adding a new one, ensure it does not conflict with
+ * PSEUDO_PARTITION_TABLE_SPCOID.
+ */
 [
 
 { oid => '1663', oid_symbol => 'DEFAULTTABLESPACE_OID',
diff --git a/src/include/catalog/pg_tablespace.h b/src/include/catalog/pg_tablespace.h
index 7816d779d8c..0e2d8051d69 100644
--- a/src/include/catalog/pg_tablespace.h
+++ b/src/include/catalog/pg_tablespace.h
@@ -21,6 +21,14 @@
 #include "catalog/genbki.h"
 #include "catalog/pg_tablespace_d.h"	/* IWYU pragma: export */
 
+/*
+ * Reserved tablespace OID for partitioned table pseudo locators.
+ * This is not an actual tablespace, just a reserved value to distinguish
+ * partitioned table statistics from regular table statistics. Ensures it does
+ * not conflict with the ones in pg_tablespace.dat.
+ */
+#define PSEUDO_PARTITION_TABLE_SPCOID 1665
+
 /* ----------------
  *		pg_tablespace definition.  cpp turns this into
  *		typedef struct FormData_pg_tablespace
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index f23dd5870da..3102f86aa24 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -17,6 +17,7 @@
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
 #include "replication/worker_internal.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -35,6 +36,12 @@
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -175,11 +182,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -669,8 +676,8 @@ extern void pgstat_init_relation(Relation rel);
 extern void pgstat_assoc_relation(Relation rel);
 extern void pgstat_unlink_relation(Relation rel);
 
-extern void pgstat_report_vacuum(Oid tableoid, bool shared,
-								 PgStat_Counter livetuples, PgStat_Counter deadtuples,
+extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
+								 PgStat_Counter deadtuples,
 								 TimestampTz starttime);
 extern void pgstat_report_analyze(Relation rel,
 								  PgStat_Counter livetuples, PgStat_Counter deadtuples,
@@ -735,8 +742,8 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index 5c1ce4d3d6a..7b24928b00d 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -764,6 +764,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index 021e2bf361f..3a9c05eaf10 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--tV2Nti+lXhoeT6I5
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v9-0002-handle-relation-statistics-correctly-during-rewri.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread

* [PATCH v9 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

This is possible as, since b14e9ce7d55c, the objoid is now uint64 and spcOid
and relNumber are 32 bits.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/access/heap/vacuumlazy.c         |   3 +-
 src/backend/postmaster/autovacuum.c          |  17 +-
 src/backend/utils/activity/pgstat_relation.c | 239 +++++++++++++++----
 src/backend/utils/adt/pgstatfuncs.c          |  22 +-
 src/include/catalog/pg_tablespace.dat        |   4 +
 src/include/catalog/pg_tablespace.h          |   8 +
 src/include/pgstat.h                         |  19 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 9 files changed, 275 insertions(+), 78 deletions(-)
   5.9% src/backend/postmaster/
  60.4% src/backend/utils/activity/
   4.9% src/backend/utils/adt/
   3.0% src/include/catalog/
   7.1% src/include/
  17.5% src/test/recovery/t/

diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c
index 62035b7f9c3..30778a15639 100644
--- a/src/backend/access/heap/vacuumlazy.c
+++ b/src/backend/access/heap/vacuumlazy.c
@@ -961,8 +961,7 @@ heap_vacuum_rel(Relation rel, const VacuumParams params,
 	 * soon in cases where the failsafe prevented significant amounts of heap
 	 * vacuuming.
 	 */
-	pgstat_report_vacuum(RelationGetRelid(rel),
-						 rel->rd_rel->relisshared,
+	pgstat_report_vacuum(rel,
 						 Max(vacrel->new_live_tuples, 0),
 						 vacrel->recently_dead_tuples +
 						 vacrel->missed_dead_tuples,
diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index 1bd3924e35e..a11174b25ad 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -2014,12 +2014,16 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		if (classForm->relkind != RELKIND_RELATION &&
 			classForm->relkind != RELKIND_MATVIEW)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * Check if it is a temp table (presumably, of some other backend's).
@@ -2048,8 +2052,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2114,6 +2117,7 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		/*
 		 * We cannot safely process other backends' temp tables, so skip 'em.
@@ -2122,6 +2126,9 @@ do_autovacuum(void)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * fetch reloptions -- if this toast table does not have them, try the
@@ -2141,8 +2148,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2939,8 +2945,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	PgStat_StatTabEntry *tabentry;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index b90754f8578..48bf93cae6e 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,37 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +197,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +215,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -207,27 +245,29 @@ pgstat_drop_relation(Relation rel)
  * Report that the table was just vacuumed and flush IO statistics.
  */
 void
-pgstat_report_vacuum(Oid tableoid, bool shared,
-					 PgStat_Counter livetuples, PgStat_Counter deadtuples,
-					 TimestampTz starttime)
+pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
+					 PgStat_Counter deadtuples, TimestampTz starttime)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
 
+	locator = rel->rd_locator;
 	/* Store the data in the table's hash table entry. */
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
 	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
-											dboid, tableoid, false);
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
+											false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -286,9 +326,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -326,9 +366,25 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -469,7 +525,16 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
+}
+
+PgStat_StatTabEntry *
+pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator)
+{
+	return (PgStat_StatTabEntry *) pgstat_fetch_entry(
+													  PGSTAT_KIND_RELATION,
+													  locator.dbOid,
+													  RelFileLocatorToPgStatObjid(locator));
 }
 
 /*
@@ -477,12 +542,14 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
+
+	return pgstat_fetch_stat_tabentry_by_locator(locator);
 }
 
 /*
@@ -504,14 +571,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -707,8 +777,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -751,7 +825,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -786,8 +860,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -921,17 +995,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1010,3 +1088,82 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index ef6fffe60b9..60ffb1679ec 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -23,13 +23,13 @@
 #include "common/ip.h"
 #include "funcapi.h"
 #include "miscadmin.h"
-#include "pgstat.h"
 #include "postmaster/bgworker.h"
 #include "replication/logicallauncher.h"
 #include "storage/proc.h"
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 
 #define UINT32_ACCESS_ONCE(var)		 ((uint32)(*((volatile uint32 *)&(var))))
@@ -1949,9 +1949,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2305,5 +2310,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/catalog/pg_tablespace.dat b/src/include/catalog/pg_tablespace.dat
index 1302a3d75cd..9430970fffd 100644
--- a/src/include/catalog/pg_tablespace.dat
+++ b/src/include/catalog/pg_tablespace.dat
@@ -10,6 +10,10 @@
 #
 #----------------------------------------------------------------------
 
+/*
+ * When adding a new one, ensure it does not conflict with
+ * PSEUDO_PARTITION_TABLE_SPCOID.
+ */
 [
 
 { oid => '1663', oid_symbol => 'DEFAULTTABLESPACE_OID',
diff --git a/src/include/catalog/pg_tablespace.h b/src/include/catalog/pg_tablespace.h
index 7816d779d8c..0e2d8051d69 100644
--- a/src/include/catalog/pg_tablespace.h
+++ b/src/include/catalog/pg_tablespace.h
@@ -21,6 +21,14 @@
 #include "catalog/genbki.h"
 #include "catalog/pg_tablespace_d.h"	/* IWYU pragma: export */
 
+/*
+ * Reserved tablespace OID for partitioned table pseudo locators.
+ * This is not an actual tablespace, just a reserved value to distinguish
+ * partitioned table statistics from regular table statistics. Ensures it does
+ * not conflict with the ones in pg_tablespace.dat.
+ */
+#define PSEUDO_PARTITION_TABLE_SPCOID 1665
+
 /* ----------------
  *		pg_tablespace definition.  cpp turns this into
  *		typedef struct FormData_pg_tablespace
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index f23dd5870da..3102f86aa24 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -17,6 +17,7 @@
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
 #include "replication/worker_internal.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -35,6 +36,12 @@
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -175,11 +182,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -669,8 +676,8 @@ extern void pgstat_init_relation(Relation rel);
 extern void pgstat_assoc_relation(Relation rel);
 extern void pgstat_unlink_relation(Relation rel);
 
-extern void pgstat_report_vacuum(Oid tableoid, bool shared,
-								 PgStat_Counter livetuples, PgStat_Counter deadtuples,
+extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
+								 PgStat_Counter deadtuples,
 								 TimestampTz starttime);
 extern void pgstat_report_analyze(Relation rel,
 								  PgStat_Counter livetuples, PgStat_Counter deadtuples,
@@ -735,8 +742,8 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index 5c1ce4d3d6a..7b24928b00d 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -764,6 +764,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index 021e2bf361f..3a9c05eaf10 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--tV2Nti+lXhoeT6I5
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v9-0002-handle-relation-statistics-correctly-during-rewri.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread

* [PATCH v14 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

This is possible as, since b14e9ce7d55c, the objoid is now uint64 and spcOid
and relNumber are 32 bits.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/postmaster/autovacuum.c          |  17 +-
 src/backend/utils/activity/pgstat_relation.c | 236 ++++++++++++++++---
 src/backend/utils/adt/pgstatfuncs.c          |  21 +-
 src/include/catalog/pg_tablespace.dat        |   4 +
 src/include/catalog/pg_tablespace.h          |   8 +
 src/include/pgstat.h                         |  15 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 8 files changed, 271 insertions(+), 71 deletions(-)
   6.1% src/backend/postmaster/
  61.7% src/backend/utils/activity/
   4.9% src/backend/utils/adt/
   3.2% src/include/catalog/
   5.6% src/include/
  18.1% src/test/recovery/t/

diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index 219673db930..4bef7ab7543 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -1992,12 +1992,16 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		if (classForm->relkind != RELKIND_RELATION &&
 			classForm->relkind != RELKIND_MATVIEW)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * Check if it is a temp table (presumably, of some other backend's).
@@ -2026,8 +2030,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2092,6 +2095,7 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		/*
 		 * We cannot safely process other backends' temp tables, so skip 'em.
@@ -2100,6 +2104,9 @@ do_autovacuum(void)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * fetch reloptions -- if this toast table does not have them, try the
@@ -2119,8 +2126,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2917,8 +2923,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	PgStat_StatTabEntry *tabentry;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index bc8c43b96aa..89bf0cbed56 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,37 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +197,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +215,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -213,20 +251,23 @@ pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
 
+	locator = rel->rd_locator;
+
 	/* Store the data in the table's hash table entry. */
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel), false);
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
+											false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -285,9 +326,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -325,9 +366,25 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -468,7 +525,16 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
+}
+
+PgStat_StatTabEntry *
+pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator)
+{
+	return (PgStat_StatTabEntry *) pgstat_fetch_entry(
+													  PGSTAT_KIND_RELATION,
+													  locator.dbOid,
+													  RelFileLocatorToPgStatObjid(locator));
 }
 
 /*
@@ -476,12 +542,14 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
+
+	return pgstat_fetch_stat_tabentry_by_locator(locator);
 }
 
 /*
@@ -503,14 +571,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -706,8 +777,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -750,7 +825,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -785,8 +860,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -920,17 +995,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1009,3 +1088,82 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index 9185a8e6b83..db882339f1e 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -30,6 +30,7 @@
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 #include "utils/tuplestore.h"
 #include "utils/wait_event.h"
@@ -2004,9 +2005,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2363,5 +2369,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/catalog/pg_tablespace.dat b/src/include/catalog/pg_tablespace.dat
index c4cde415219..73ed046be31 100644
--- a/src/include/catalog/pg_tablespace.dat
+++ b/src/include/catalog/pg_tablespace.dat
@@ -10,6 +10,10 @@
 #
 #----------------------------------------------------------------------
 
+/*
+ * When adding a new one, ensure it does not conflict with
+ * PSEUDO_PARTITION_TABLE_SPCOID.
+ */
 [
 
 { oid => '1663', oid_symbol => 'DEFAULTTABLESPACE_OID',
diff --git a/src/include/catalog/pg_tablespace.h b/src/include/catalog/pg_tablespace.h
index 3bd4a74f003..e5e818eb0c5 100644
--- a/src/include/catalog/pg_tablespace.h
+++ b/src/include/catalog/pg_tablespace.h
@@ -21,6 +21,14 @@
 #include "catalog/genbki.h"
 #include "catalog/pg_tablespace_d.h"	/* IWYU pragma: export */
 
+/*
+ * Reserved tablespace OID for partitioned table pseudo locators.
+ * This is not an actual tablespace, just a reserved value to distinguish
+ * partitioned table statistics from regular table statistics. Ensures it does
+ * not conflict with the ones in pg_tablespace.dat.
+ */
+#define PSEUDO_PARTITION_TABLE_SPCOID 1665
+
 /* ----------------
  *		pg_tablespace definition.  cpp turns this into
  *		typedef struct FormData_pg_tablespace
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index 8e3549c3752..ef740aff26e 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -16,6 +16,7 @@
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
 #include "storage/locktag.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -39,6 +40,12 @@ typedef struct RelationData *Relation;
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -179,11 +186,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -766,8 +773,8 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index 97704421a92..e34b008e893 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -786,6 +786,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index cdc427dbc78..4d00087dc6f 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--96iU9DKeu+W+YlSa
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v14-0002-handle-relation-statistics-correctly-during-rewr.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread

* [PATCH v10 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

This is possible as, since b14e9ce7d55c, the objoid is now uint64 and spcOid
and relNumber are 32 bits.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/postmaster/autovacuum.c          |  17 +-
 src/backend/utils/activity/pgstat_relation.c | 236 ++++++++++++++++---
 src/backend/utils/adt/pgstatfuncs.c          |  22 +-
 src/include/catalog/pg_tablespace.dat        |   4 +
 src/include/catalog/pg_tablespace.h          |   8 +
 src/include/pgstat.h                         |  15 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 8 files changed, 271 insertions(+), 72 deletions(-)
   6.1% src/backend/postmaster/
  61.6% src/backend/utils/activity/
   5.1% src/backend/utils/adt/
   3.2% src/include/catalog/
   5.6% src/include/
  18.1% src/test/recovery/t/

diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index 1bd3924e35e..a11174b25ad 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -2014,12 +2014,16 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		if (classForm->relkind != RELKIND_RELATION &&
 			classForm->relkind != RELKIND_MATVIEW)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * Check if it is a temp table (presumably, of some other backend's).
@@ -2048,8 +2052,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2114,6 +2117,7 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		/*
 		 * We cannot safely process other backends' temp tables, so skip 'em.
@@ -2122,6 +2126,9 @@ do_autovacuum(void)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * fetch reloptions -- if this toast table does not have them, try the
@@ -2141,8 +2148,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2939,8 +2945,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	PgStat_StatTabEntry *tabentry;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index 55a10c299db..a267e93f8be 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,37 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +197,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +215,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -213,20 +251,23 @@ pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
 
+	locator = rel->rd_locator;
+
 	/* Store the data in the table's hash table entry. */
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel), false);
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
+											false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -285,9 +326,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -325,9 +366,25 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -468,7 +525,16 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
+}
+
+PgStat_StatTabEntry *
+pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator)
+{
+	return (PgStat_StatTabEntry *) pgstat_fetch_entry(
+													  PGSTAT_KIND_RELATION,
+													  locator.dbOid,
+													  RelFileLocatorToPgStatObjid(locator));
 }
 
 /*
@@ -476,12 +542,14 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
+
+	return pgstat_fetch_stat_tabentry_by_locator(locator);
 }
 
 /*
@@ -503,14 +571,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -706,8 +777,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -750,7 +825,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -785,8 +860,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -920,17 +995,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1009,3 +1088,82 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index ef6fffe60b9..60ffb1679ec 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -23,13 +23,13 @@
 #include "common/ip.h"
 #include "funcapi.h"
 #include "miscadmin.h"
-#include "pgstat.h"
 #include "postmaster/bgworker.h"
 #include "replication/logicallauncher.h"
 #include "storage/proc.h"
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 
 #define UINT32_ACCESS_ONCE(var)		 ((uint32)(*((volatile uint32 *)&(var))))
@@ -1949,9 +1949,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2305,5 +2310,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/catalog/pg_tablespace.dat b/src/include/catalog/pg_tablespace.dat
index 1302a3d75cd..9430970fffd 100644
--- a/src/include/catalog/pg_tablespace.dat
+++ b/src/include/catalog/pg_tablespace.dat
@@ -10,6 +10,10 @@
 #
 #----------------------------------------------------------------------
 
+/*
+ * When adding a new one, ensure it does not conflict with
+ * PSEUDO_PARTITION_TABLE_SPCOID.
+ */
 [
 
 { oid => '1663', oid_symbol => 'DEFAULTTABLESPACE_OID',
diff --git a/src/include/catalog/pg_tablespace.h b/src/include/catalog/pg_tablespace.h
index 7816d779d8c..0e2d8051d69 100644
--- a/src/include/catalog/pg_tablespace.h
+++ b/src/include/catalog/pg_tablespace.h
@@ -21,6 +21,14 @@
 #include "catalog/genbki.h"
 #include "catalog/pg_tablespace_d.h"	/* IWYU pragma: export */
 
+/*
+ * Reserved tablespace OID for partitioned table pseudo locators.
+ * This is not an actual tablespace, just a reserved value to distinguish
+ * partitioned table statistics from regular table statistics. Ensures it does
+ * not conflict with the ones in pg_tablespace.dat.
+ */
+#define PSEUDO_PARTITION_TABLE_SPCOID 1665
+
 /* ----------------
  *		pg_tablespace definition.  cpp turns this into
  *		typedef struct FormData_pg_tablespace
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index 6714363144a..3102f86aa24 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -17,6 +17,7 @@
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
 #include "replication/worker_internal.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -35,6 +36,12 @@
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -175,11 +182,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -735,8 +742,8 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index 5c1ce4d3d6a..7b24928b00d 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -764,6 +764,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index 021e2bf361f..3a9c05eaf10 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--xJsrX25x3tm7Tn81
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v10-0002-handle-relation-statistics-correctly-during-rewr.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread

* [PATCH v11 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

This is possible as, since b14e9ce7d55c, the objoid is now uint64 and spcOid
and relNumber are 32 bits.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/postmaster/autovacuum.c          |  17 +-
 src/backend/utils/activity/pgstat_relation.c | 236 ++++++++++++++++---
 src/backend/utils/adt/pgstatfuncs.c          |  22 +-
 src/include/catalog/pg_tablespace.dat        |   4 +
 src/include/catalog/pg_tablespace.h          |   8 +
 src/include/pgstat.h                         |  15 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 8 files changed, 271 insertions(+), 72 deletions(-)
   6.1% src/backend/postmaster/
  61.6% src/backend/utils/activity/
   5.1% src/backend/utils/adt/
   3.2% src/include/catalog/
   5.6% src/include/
  18.1% src/test/recovery/t/

diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index 6fde740465f..4bd5a6824bd 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -1990,12 +1990,16 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		if (classForm->relkind != RELKIND_RELATION &&
 			classForm->relkind != RELKIND_MATVIEW)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * Check if it is a temp table (presumably, of some other backend's).
@@ -2024,8 +2028,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2090,6 +2093,7 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		/*
 		 * We cannot safely process other backends' temp tables, so skip 'em.
@@ -2098,6 +2102,9 @@ do_autovacuum(void)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * fetch reloptions -- if this toast table does not have them, try the
@@ -2117,8 +2124,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2915,8 +2921,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	PgStat_StatTabEntry *tabentry;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index bc8c43b96aa..89bf0cbed56 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,37 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +197,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +215,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -213,20 +251,23 @@ pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
 
+	locator = rel->rd_locator;
+
 	/* Store the data in the table's hash table entry. */
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel), false);
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
+											false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -285,9 +326,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -325,9 +366,25 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -468,7 +525,16 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
+}
+
+PgStat_StatTabEntry *
+pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator)
+{
+	return (PgStat_StatTabEntry *) pgstat_fetch_entry(
+													  PGSTAT_KIND_RELATION,
+													  locator.dbOid,
+													  RelFileLocatorToPgStatObjid(locator));
 }
 
 /*
@@ -476,12 +542,14 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
+
+	return pgstat_fetch_stat_tabentry_by_locator(locator);
 }
 
 /*
@@ -503,14 +571,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -706,8 +777,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -750,7 +825,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -785,8 +860,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -920,17 +995,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1009,3 +1088,82 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index b1df96e7b0b..a5a00d6f018 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -23,13 +23,13 @@
 #include "common/ip.h"
 #include "funcapi.h"
 #include "miscadmin.h"
-#include "pgstat.h"
 #include "postmaster/bgworker.h"
 #include "replication/logicallauncher.h"
 #include "storage/proc.h"
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 
 #define UINT32_ACCESS_ONCE(var)		 ((uint32)(*((volatile uint32 *)&(var))))
@@ -1961,9 +1961,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2317,5 +2322,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/catalog/pg_tablespace.dat b/src/include/catalog/pg_tablespace.dat
index c4cde415219..73ed046be31 100644
--- a/src/include/catalog/pg_tablespace.dat
+++ b/src/include/catalog/pg_tablespace.dat
@@ -10,6 +10,10 @@
 #
 #----------------------------------------------------------------------
 
+/*
+ * When adding a new one, ensure it does not conflict with
+ * PSEUDO_PARTITION_TABLE_SPCOID.
+ */
 [
 
 { oid => '1663', oid_symbol => 'DEFAULTTABLESPACE_OID',
diff --git a/src/include/catalog/pg_tablespace.h b/src/include/catalog/pg_tablespace.h
index fe7a5ab538f..f511adc3965 100644
--- a/src/include/catalog/pg_tablespace.h
+++ b/src/include/catalog/pg_tablespace.h
@@ -21,6 +21,14 @@
 #include "catalog/genbki.h"
 #include "catalog/pg_tablespace_d.h"	/* IWYU pragma: export */
 
+/*
+ * Reserved tablespace OID for partitioned table pseudo locators.
+ * This is not an actual tablespace, just a reserved value to distinguish
+ * partitioned table statistics from regular table statistics. Ensures it does
+ * not conflict with the ones in pg_tablespace.dat.
+ */
+#define PSEUDO_PARTITION_TABLE_SPCOID 1665
+
 /* ----------------
  *		pg_tablespace definition.  cpp turns this into
  *		typedef struct FormData_pg_tablespace
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index 9bb777c3d5a..e9e4d32c3b8 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -16,6 +16,7 @@
 #include "portability/instr_time.h"
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -34,6 +35,12 @@
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -174,11 +181,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -734,8 +741,8 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index 9b8fbae00ed..42f29496551 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -765,6 +765,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index cdc427dbc78..4d00087dc6f 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--zLA8DXafBEP889wv
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v11-0002-handle-relation-statistics-correctly-during-rewr.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread

* [PATCH v15 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

This is possible as, since b14e9ce7d55c, the objoid is now uint64 and spcOid
and relNumber are 32 bits.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/postmaster/autovacuum.c          |  18 +-
 src/backend/utils/activity/pgstat_relation.c | 236 ++++++++++++++++---
 src/backend/utils/adt/pgstatfuncs.c          |  21 +-
 src/include/catalog/pg_tablespace.dat        |   4 +
 src/include/catalog/pg_tablespace.h          |   8 +
 src/include/pgstat.h                         |  15 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 8 files changed, 272 insertions(+), 71 deletions(-)
   6.2% src/backend/postmaster/
  61.7% src/backend/utils/activity/
   4.9% src/backend/utils/adt/
   3.2% src/include/catalog/
   5.6% src/include/
  18.1% src/test/recovery/t/

diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index d695f1de4bd..0a028b18d66 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -2035,6 +2035,7 @@ do_autovacuum(void)
 		bool		doanalyze;
 		bool		wraparound;
 		AutoVacuumScores scores;
+		RelFileLocator locator;
 
 		if (classForm->relkind != RELKIND_RELATION &&
 			classForm->relkind != RELKIND_MATVIEW)
@@ -2042,6 +2043,10 @@ do_autovacuum(void)
 
 		relid = classForm->oid;
 
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
+
 		/*
 		 * Check if it is a temp table (presumably, of some other backend's).
 		 * We cannot safely process other backends' temp tables.
@@ -2069,8 +2074,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2143,6 +2147,7 @@ do_autovacuum(void)
 		bool		doanalyze;
 		bool		wraparound;
 		AutoVacuumScores scores;
+		RelFileLocator locator;
 
 		/*
 		 * We cannot safely process other backends' temp tables, so skip 'em.
@@ -2151,6 +2156,9 @@ do_autovacuum(void)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * fetch reloptions -- if this toast table does not have them, try the
@@ -2170,8 +2178,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2988,8 +2995,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	AutoVacuumScores scores;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index bc8c43b96aa..89bf0cbed56 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,37 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +197,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +215,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -213,20 +251,23 @@ pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
 
+	locator = rel->rd_locator;
+
 	/* Store the data in the table's hash table entry. */
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel), false);
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
+											false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -285,9 +326,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -325,9 +366,25 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -468,7 +525,16 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
+}
+
+PgStat_StatTabEntry *
+pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator)
+{
+	return (PgStat_StatTabEntry *) pgstat_fetch_entry(
+													  PGSTAT_KIND_RELATION,
+													  locator.dbOid,
+													  RelFileLocatorToPgStatObjid(locator));
 }
 
 /*
@@ -476,12 +542,14 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
+
+	return pgstat_fetch_stat_tabentry_by_locator(locator);
 }
 
 /*
@@ -503,14 +571,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -706,8 +777,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -750,7 +825,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -785,8 +860,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -920,17 +995,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1009,3 +1088,82 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index 9185a8e6b83..db882339f1e 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -30,6 +30,7 @@
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 #include "utils/tuplestore.h"
 #include "utils/wait_event.h"
@@ -2004,9 +2005,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2363,5 +2369,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/catalog/pg_tablespace.dat b/src/include/catalog/pg_tablespace.dat
index c4cde415219..73ed046be31 100644
--- a/src/include/catalog/pg_tablespace.dat
+++ b/src/include/catalog/pg_tablespace.dat
@@ -10,6 +10,10 @@
 #
 #----------------------------------------------------------------------
 
+/*
+ * When adding a new one, ensure it does not conflict with
+ * PSEUDO_PARTITION_TABLE_SPCOID.
+ */
 [
 
 { oid => '1663', oid_symbol => 'DEFAULTTABLESPACE_OID',
diff --git a/src/include/catalog/pg_tablespace.h b/src/include/catalog/pg_tablespace.h
index 3bd4a74f003..e5e818eb0c5 100644
--- a/src/include/catalog/pg_tablespace.h
+++ b/src/include/catalog/pg_tablespace.h
@@ -21,6 +21,14 @@
 #include "catalog/genbki.h"
 #include "catalog/pg_tablespace_d.h"	/* IWYU pragma: export */
 
+/*
+ * Reserved tablespace OID for partitioned table pseudo locators.
+ * This is not an actual tablespace, just a reserved value to distinguish
+ * partitioned table statistics from regular table statistics. Ensures it does
+ * not conflict with the ones in pg_tablespace.dat.
+ */
+#define PSEUDO_PARTITION_TABLE_SPCOID 1665
+
 /* ----------------
  *		pg_tablespace definition.  cpp turns this into
  *		typedef struct FormData_pg_tablespace
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index 8e3549c3752..ef740aff26e 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -16,6 +16,7 @@
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
 #include "storage/locktag.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -39,6 +40,12 @@ typedef struct RelationData *Relation;
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -179,11 +186,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -766,8 +773,8 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index eed4c6b359c..1090f049058 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -786,6 +786,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index cdc427dbc78..4d00087dc6f 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--Og8T0Lf2JVsuKOzn
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v15-0002-handle-relation-statistics-correctly-during-rewr.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread

* [PATCH v8 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/access/heap/vacuumlazy.c         |   3 +-
 src/backend/postmaster/autovacuum.c          |   9 +-
 src/backend/utils/activity/pgstat_relation.c | 234 +++++++++++++++----
 src/backend/utils/adt/pgstatfuncs.c          |  22 +-
 src/include/pgstat.h                         |  18 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 7 files changed, 249 insertions(+), 78 deletions(-)
   3.3% src/backend/postmaster/
  64.6% src/backend/utils/activity/
   5.3% src/backend/utils/adt/
   7.0% src/include/
  18.7% src/test/recovery/t/

diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c
index 62035b7f9c3..a9b2b4e1033 100644
--- a/src/backend/access/heap/vacuumlazy.c
+++ b/src/backend/access/heap/vacuumlazy.c
@@ -961,8 +961,7 @@ heap_vacuum_rel(Relation rel, const VacuumParams params,
 	 * soon in cases where the failsafe prevented significant amounts of heap
 	 * vacuuming.
 	 */
-	pgstat_report_vacuum(RelationGetRelid(rel),
-						 rel->rd_rel->relisshared,
+	pgstat_report_vacuum(rel->rd_locator,
 						 Max(vacrel->new_live_tuples, 0),
 						 vacrel->recently_dead_tuples +
 						 vacrel->missed_dead_tuples,
diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index 1bd3924e35e..563a3697690 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -2048,8 +2048,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2141,8 +2140,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2939,8 +2937,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	PgStat_StatTabEntry *tabentry;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index b90754f8578..ee6f2eb2bdb 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,38 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use the relation OID as relNumber.
+		 * No collision with regular relations is possible because relNumbers
+		 * are also assigned from the pg_class OID space (see
+		 * GetNewRelFileNumber()), making each value unique across the
+		 * database regardless of spcOid.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = InvalidOid;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +198,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +216,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -207,14 +246,12 @@ pgstat_drop_relation(Relation rel)
  * Report that the table was just vacuumed and flush IO statistics.
  */
 void
-pgstat_report_vacuum(Oid tableoid, bool shared,
-					 PgStat_Counter livetuples, PgStat_Counter deadtuples,
-					 TimestampTz starttime)
+pgstat_report_vacuum(RelFileLocator locator, PgStat_Counter livetuples,
+					 PgStat_Counter deadtuples, TimestampTz starttime)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
 
@@ -227,7 +264,7 @@ pgstat_report_vacuum(Oid tableoid, bool shared,
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
 	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
-											dboid, tableoid, false);
+											locator.dbOid, RelFileLocatorToPgStatObjid(locator), false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -286,9 +323,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -326,9 +363,26 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use the relation OID as relNumber.
+		 * No collision with regular relations is possible because relNumbers
+		 * are also assigned from the pg_class OID space (see
+		 * GetNewRelFileNumber()), making each value unique across the
+		 * database regardless of spcOid.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = InvalidOid;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -469,7 +523,7 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
 }
 
 /*
@@ -477,12 +531,19 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	PgStat_StatTabEntry *tabentry;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	/* fetch the stats entry using the relfilenode based key */
+	tabentry = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION,
+														  locator.dbOid,
+														  RelFileLocatorToPgStatObjid(locator));
+	return tabentry;
 }
 
 /*
@@ -504,14 +565,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -707,8 +771,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -751,7 +819,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -786,8 +854,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -921,17 +989,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1010,3 +1082,83 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use the relation OID as relNumber.
+		 * No collision with regular relations is possible because relNumbers
+		 * are also assigned from the pg_class OID space (see
+		 * GetNewRelFileNumber()), making each value unique across the
+		 * database regardless of spcOid.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = InvalidOid;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index ef6fffe60b9..60ffb1679ec 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -23,13 +23,13 @@
 #include "common/ip.h"
 #include "funcapi.h"
 #include "miscadmin.h"
-#include "pgstat.h"
 #include "postmaster/bgworker.h"
 #include "replication/logicallauncher.h"
 #include "storage/proc.h"
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 
 #define UINT32_ACCESS_ONCE(var)		 ((uint32)(*((volatile uint32 *)&(var))))
@@ -1949,9 +1949,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2305,5 +2310,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index f23dd5870da..c9e451094f3 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -17,6 +17,7 @@
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
 #include "replication/worker_internal.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -35,6 +36,12 @@
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -175,11 +182,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -669,8 +676,8 @@ extern void pgstat_init_relation(Relation rel);
 extern void pgstat_assoc_relation(Relation rel);
 extern void pgstat_unlink_relation(Relation rel);
 
-extern void pgstat_report_vacuum(Oid tableoid, bool shared,
-								 PgStat_Counter livetuples, PgStat_Counter deadtuples,
+extern void pgstat_report_vacuum(RelFileLocator locator, PgStat_Counter livetuples,
+								 PgStat_Counter deadtuples,
 								 TimestampTz starttime);
 extern void pgstat_report_analyze(Relation rel,
 								  PgStat_Counter livetuples, PgStat_Counter deadtuples,
@@ -735,8 +742,7 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index 5c1ce4d3d6a..7b24928b00d 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -764,6 +764,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index 021e2bf361f..3a9c05eaf10 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--BHD/2LBqtxMXWADJ
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v8-0002-handle-relation-statistics-correctly-during-rewri.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread

* [PATCH v11 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

This is possible as, since b14e9ce7d55c, the objoid is now uint64 and spcOid
and relNumber are 32 bits.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/postmaster/autovacuum.c          |  17 +-
 src/backend/utils/activity/pgstat_relation.c | 236 ++++++++++++++++---
 src/backend/utils/adt/pgstatfuncs.c          |  22 +-
 src/include/catalog/pg_tablespace.dat        |   4 +
 src/include/catalog/pg_tablespace.h          |   8 +
 src/include/pgstat.h                         |  15 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 8 files changed, 271 insertions(+), 72 deletions(-)
   6.1% src/backend/postmaster/
  61.6% src/backend/utils/activity/
   5.1% src/backend/utils/adt/
   3.2% src/include/catalog/
   5.6% src/include/
  18.1% src/test/recovery/t/

diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index 6fde740465f..4bd5a6824bd 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -1990,12 +1990,16 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		if (classForm->relkind != RELKIND_RELATION &&
 			classForm->relkind != RELKIND_MATVIEW)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * Check if it is a temp table (presumably, of some other backend's).
@@ -2024,8 +2028,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2090,6 +2093,7 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		/*
 		 * We cannot safely process other backends' temp tables, so skip 'em.
@@ -2098,6 +2102,9 @@ do_autovacuum(void)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * fetch reloptions -- if this toast table does not have them, try the
@@ -2117,8 +2124,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2915,8 +2921,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	PgStat_StatTabEntry *tabentry;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index bc8c43b96aa..89bf0cbed56 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,37 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +197,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +215,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -213,20 +251,23 @@ pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
 
+	locator = rel->rd_locator;
+
 	/* Store the data in the table's hash table entry. */
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel), false);
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
+											false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -285,9 +326,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -325,9 +366,25 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -468,7 +525,16 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
+}
+
+PgStat_StatTabEntry *
+pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator)
+{
+	return (PgStat_StatTabEntry *) pgstat_fetch_entry(
+													  PGSTAT_KIND_RELATION,
+													  locator.dbOid,
+													  RelFileLocatorToPgStatObjid(locator));
 }
 
 /*
@@ -476,12 +542,14 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
+
+	return pgstat_fetch_stat_tabentry_by_locator(locator);
 }
 
 /*
@@ -503,14 +571,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -706,8 +777,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -750,7 +825,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -785,8 +860,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -920,17 +995,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1009,3 +1088,82 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index b1df96e7b0b..a5a00d6f018 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -23,13 +23,13 @@
 #include "common/ip.h"
 #include "funcapi.h"
 #include "miscadmin.h"
-#include "pgstat.h"
 #include "postmaster/bgworker.h"
 #include "replication/logicallauncher.h"
 #include "storage/proc.h"
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 
 #define UINT32_ACCESS_ONCE(var)		 ((uint32)(*((volatile uint32 *)&(var))))
@@ -1961,9 +1961,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2317,5 +2322,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/catalog/pg_tablespace.dat b/src/include/catalog/pg_tablespace.dat
index c4cde415219..73ed046be31 100644
--- a/src/include/catalog/pg_tablespace.dat
+++ b/src/include/catalog/pg_tablespace.dat
@@ -10,6 +10,10 @@
 #
 #----------------------------------------------------------------------
 
+/*
+ * When adding a new one, ensure it does not conflict with
+ * PSEUDO_PARTITION_TABLE_SPCOID.
+ */
 [
 
 { oid => '1663', oid_symbol => 'DEFAULTTABLESPACE_OID',
diff --git a/src/include/catalog/pg_tablespace.h b/src/include/catalog/pg_tablespace.h
index fe7a5ab538f..f511adc3965 100644
--- a/src/include/catalog/pg_tablespace.h
+++ b/src/include/catalog/pg_tablespace.h
@@ -21,6 +21,14 @@
 #include "catalog/genbki.h"
 #include "catalog/pg_tablespace_d.h"	/* IWYU pragma: export */
 
+/*
+ * Reserved tablespace OID for partitioned table pseudo locators.
+ * This is not an actual tablespace, just a reserved value to distinguish
+ * partitioned table statistics from regular table statistics. Ensures it does
+ * not conflict with the ones in pg_tablespace.dat.
+ */
+#define PSEUDO_PARTITION_TABLE_SPCOID 1665
+
 /* ----------------
  *		pg_tablespace definition.  cpp turns this into
  *		typedef struct FormData_pg_tablespace
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index 9bb777c3d5a..e9e4d32c3b8 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -16,6 +16,7 @@
 #include "portability/instr_time.h"
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -34,6 +35,12 @@
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -174,11 +181,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -734,8 +741,8 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index 9b8fbae00ed..42f29496551 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -765,6 +765,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index cdc427dbc78..4d00087dc6f 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--zLA8DXafBEP889wv
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v11-0002-handle-relation-statistics-correctly-during-rewr.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread

* [PATCH v11 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

This is possible as, since b14e9ce7d55c, the objoid is now uint64 and spcOid
and relNumber are 32 bits.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/postmaster/autovacuum.c          |  17 +-
 src/backend/utils/activity/pgstat_relation.c | 236 ++++++++++++++++---
 src/backend/utils/adt/pgstatfuncs.c          |  22 +-
 src/include/catalog/pg_tablespace.dat        |   4 +
 src/include/catalog/pg_tablespace.h          |   8 +
 src/include/pgstat.h                         |  15 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 8 files changed, 271 insertions(+), 72 deletions(-)
   6.1% src/backend/postmaster/
  61.6% src/backend/utils/activity/
   5.1% src/backend/utils/adt/
   3.2% src/include/catalog/
   5.6% src/include/
  18.1% src/test/recovery/t/

diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index 6fde740465f..4bd5a6824bd 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -1990,12 +1990,16 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		if (classForm->relkind != RELKIND_RELATION &&
 			classForm->relkind != RELKIND_MATVIEW)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * Check if it is a temp table (presumably, of some other backend's).
@@ -2024,8 +2028,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2090,6 +2093,7 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		/*
 		 * We cannot safely process other backends' temp tables, so skip 'em.
@@ -2098,6 +2102,9 @@ do_autovacuum(void)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * fetch reloptions -- if this toast table does not have them, try the
@@ -2117,8 +2124,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2915,8 +2921,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	PgStat_StatTabEntry *tabentry;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index bc8c43b96aa..89bf0cbed56 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,37 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +197,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +215,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -213,20 +251,23 @@ pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
 
+	locator = rel->rd_locator;
+
 	/* Store the data in the table's hash table entry. */
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel), false);
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
+											false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -285,9 +326,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -325,9 +366,25 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -468,7 +525,16 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
+}
+
+PgStat_StatTabEntry *
+pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator)
+{
+	return (PgStat_StatTabEntry *) pgstat_fetch_entry(
+													  PGSTAT_KIND_RELATION,
+													  locator.dbOid,
+													  RelFileLocatorToPgStatObjid(locator));
 }
 
 /*
@@ -476,12 +542,14 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
+
+	return pgstat_fetch_stat_tabentry_by_locator(locator);
 }
 
 /*
@@ -503,14 +571,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -706,8 +777,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -750,7 +825,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -785,8 +860,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -920,17 +995,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1009,3 +1088,82 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index b1df96e7b0b..a5a00d6f018 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -23,13 +23,13 @@
 #include "common/ip.h"
 #include "funcapi.h"
 #include "miscadmin.h"
-#include "pgstat.h"
 #include "postmaster/bgworker.h"
 #include "replication/logicallauncher.h"
 #include "storage/proc.h"
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 
 #define UINT32_ACCESS_ONCE(var)		 ((uint32)(*((volatile uint32 *)&(var))))
@@ -1961,9 +1961,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2317,5 +2322,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/catalog/pg_tablespace.dat b/src/include/catalog/pg_tablespace.dat
index c4cde415219..73ed046be31 100644
--- a/src/include/catalog/pg_tablespace.dat
+++ b/src/include/catalog/pg_tablespace.dat
@@ -10,6 +10,10 @@
 #
 #----------------------------------------------------------------------
 
+/*
+ * When adding a new one, ensure it does not conflict with
+ * PSEUDO_PARTITION_TABLE_SPCOID.
+ */
 [
 
 { oid => '1663', oid_symbol => 'DEFAULTTABLESPACE_OID',
diff --git a/src/include/catalog/pg_tablespace.h b/src/include/catalog/pg_tablespace.h
index fe7a5ab538f..f511adc3965 100644
--- a/src/include/catalog/pg_tablespace.h
+++ b/src/include/catalog/pg_tablespace.h
@@ -21,6 +21,14 @@
 #include "catalog/genbki.h"
 #include "catalog/pg_tablespace_d.h"	/* IWYU pragma: export */
 
+/*
+ * Reserved tablespace OID for partitioned table pseudo locators.
+ * This is not an actual tablespace, just a reserved value to distinguish
+ * partitioned table statistics from regular table statistics. Ensures it does
+ * not conflict with the ones in pg_tablespace.dat.
+ */
+#define PSEUDO_PARTITION_TABLE_SPCOID 1665
+
 /* ----------------
  *		pg_tablespace definition.  cpp turns this into
  *		typedef struct FormData_pg_tablespace
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index 9bb777c3d5a..e9e4d32c3b8 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -16,6 +16,7 @@
 #include "portability/instr_time.h"
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -34,6 +35,12 @@
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -174,11 +181,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -734,8 +741,8 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index 9b8fbae00ed..42f29496551 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -765,6 +765,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index cdc427dbc78..4d00087dc6f 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--zLA8DXafBEP889wv
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v11-0002-handle-relation-statistics-correctly-during-rewr.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread

* [PATCH v9 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

This is possible as, since b14e9ce7d55c, the objoid is now uint64 and spcOid
and relNumber are 32 bits.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/access/heap/vacuumlazy.c         |   3 +-
 src/backend/postmaster/autovacuum.c          |  17 +-
 src/backend/utils/activity/pgstat_relation.c | 239 +++++++++++++++----
 src/backend/utils/adt/pgstatfuncs.c          |  22 +-
 src/include/catalog/pg_tablespace.dat        |   4 +
 src/include/catalog/pg_tablespace.h          |   8 +
 src/include/pgstat.h                         |  19 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 9 files changed, 275 insertions(+), 78 deletions(-)
   5.9% src/backend/postmaster/
  60.4% src/backend/utils/activity/
   4.9% src/backend/utils/adt/
   3.0% src/include/catalog/
   7.1% src/include/
  17.5% src/test/recovery/t/

diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c
index 62035b7f9c3..30778a15639 100644
--- a/src/backend/access/heap/vacuumlazy.c
+++ b/src/backend/access/heap/vacuumlazy.c
@@ -961,8 +961,7 @@ heap_vacuum_rel(Relation rel, const VacuumParams params,
 	 * soon in cases where the failsafe prevented significant amounts of heap
 	 * vacuuming.
 	 */
-	pgstat_report_vacuum(RelationGetRelid(rel),
-						 rel->rd_rel->relisshared,
+	pgstat_report_vacuum(rel,
 						 Max(vacrel->new_live_tuples, 0),
 						 vacrel->recently_dead_tuples +
 						 vacrel->missed_dead_tuples,
diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index 1bd3924e35e..a11174b25ad 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -2014,12 +2014,16 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		if (classForm->relkind != RELKIND_RELATION &&
 			classForm->relkind != RELKIND_MATVIEW)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * Check if it is a temp table (presumably, of some other backend's).
@@ -2048,8 +2052,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2114,6 +2117,7 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		/*
 		 * We cannot safely process other backends' temp tables, so skip 'em.
@@ -2122,6 +2126,9 @@ do_autovacuum(void)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * fetch reloptions -- if this toast table does not have them, try the
@@ -2141,8 +2148,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2939,8 +2945,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	PgStat_StatTabEntry *tabentry;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index b90754f8578..48bf93cae6e 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,37 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +197,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +215,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -207,27 +245,29 @@ pgstat_drop_relation(Relation rel)
  * Report that the table was just vacuumed and flush IO statistics.
  */
 void
-pgstat_report_vacuum(Oid tableoid, bool shared,
-					 PgStat_Counter livetuples, PgStat_Counter deadtuples,
-					 TimestampTz starttime)
+pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
+					 PgStat_Counter deadtuples, TimestampTz starttime)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
 
+	locator = rel->rd_locator;
 	/* Store the data in the table's hash table entry. */
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
 	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
-											dboid, tableoid, false);
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
+											false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -286,9 +326,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -326,9 +366,25 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -469,7 +525,16 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
+}
+
+PgStat_StatTabEntry *
+pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator)
+{
+	return (PgStat_StatTabEntry *) pgstat_fetch_entry(
+													  PGSTAT_KIND_RELATION,
+													  locator.dbOid,
+													  RelFileLocatorToPgStatObjid(locator));
 }
 
 /*
@@ -477,12 +542,14 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
+
+	return pgstat_fetch_stat_tabentry_by_locator(locator);
 }
 
 /*
@@ -504,14 +571,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -707,8 +777,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -751,7 +825,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -786,8 +860,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -921,17 +995,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1010,3 +1088,82 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index ef6fffe60b9..60ffb1679ec 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -23,13 +23,13 @@
 #include "common/ip.h"
 #include "funcapi.h"
 #include "miscadmin.h"
-#include "pgstat.h"
 #include "postmaster/bgworker.h"
 #include "replication/logicallauncher.h"
 #include "storage/proc.h"
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 
 #define UINT32_ACCESS_ONCE(var)		 ((uint32)(*((volatile uint32 *)&(var))))
@@ -1949,9 +1949,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2305,5 +2310,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/catalog/pg_tablespace.dat b/src/include/catalog/pg_tablespace.dat
index 1302a3d75cd..9430970fffd 100644
--- a/src/include/catalog/pg_tablespace.dat
+++ b/src/include/catalog/pg_tablespace.dat
@@ -10,6 +10,10 @@
 #
 #----------------------------------------------------------------------
 
+/*
+ * When adding a new one, ensure it does not conflict with
+ * PSEUDO_PARTITION_TABLE_SPCOID.
+ */
 [
 
 { oid => '1663', oid_symbol => 'DEFAULTTABLESPACE_OID',
diff --git a/src/include/catalog/pg_tablespace.h b/src/include/catalog/pg_tablespace.h
index 7816d779d8c..0e2d8051d69 100644
--- a/src/include/catalog/pg_tablespace.h
+++ b/src/include/catalog/pg_tablespace.h
@@ -21,6 +21,14 @@
 #include "catalog/genbki.h"
 #include "catalog/pg_tablespace_d.h"	/* IWYU pragma: export */
 
+/*
+ * Reserved tablespace OID for partitioned table pseudo locators.
+ * This is not an actual tablespace, just a reserved value to distinguish
+ * partitioned table statistics from regular table statistics. Ensures it does
+ * not conflict with the ones in pg_tablespace.dat.
+ */
+#define PSEUDO_PARTITION_TABLE_SPCOID 1665
+
 /* ----------------
  *		pg_tablespace definition.  cpp turns this into
  *		typedef struct FormData_pg_tablespace
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index f23dd5870da..3102f86aa24 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -17,6 +17,7 @@
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
 #include "replication/worker_internal.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -35,6 +36,12 @@
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -175,11 +182,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -669,8 +676,8 @@ extern void pgstat_init_relation(Relation rel);
 extern void pgstat_assoc_relation(Relation rel);
 extern void pgstat_unlink_relation(Relation rel);
 
-extern void pgstat_report_vacuum(Oid tableoid, bool shared,
-								 PgStat_Counter livetuples, PgStat_Counter deadtuples,
+extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
+								 PgStat_Counter deadtuples,
 								 TimestampTz starttime);
 extern void pgstat_report_analyze(Relation rel,
 								  PgStat_Counter livetuples, PgStat_Counter deadtuples,
@@ -735,8 +742,8 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index 5c1ce4d3d6a..7b24928b00d 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -764,6 +764,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index 021e2bf361f..3a9c05eaf10 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--tV2Nti+lXhoeT6I5
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v9-0002-handle-relation-statistics-correctly-during-rewri.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread

* [PATCH v12 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

This is possible as, since b14e9ce7d55c, the objoid is now uint64 and spcOid
and relNumber are 32 bits.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/postmaster/autovacuum.c          |  17 +-
 src/backend/utils/activity/pgstat_relation.c | 236 ++++++++++++++++---
 src/backend/utils/adt/pgstatfuncs.c          |  22 +-
 src/include/catalog/pg_tablespace.dat        |   4 +
 src/include/catalog/pg_tablespace.h          |   8 +
 src/include/pgstat.h                         |  15 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 8 files changed, 271 insertions(+), 72 deletions(-)
   6.1% src/backend/postmaster/
  61.6% src/backend/utils/activity/
   5.1% src/backend/utils/adt/
   3.2% src/include/catalog/
   5.6% src/include/
  18.1% src/test/recovery/t/

diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index 695e187ba11..5d80e0bd212 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -1991,12 +1991,16 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		if (classForm->relkind != RELKIND_RELATION &&
 			classForm->relkind != RELKIND_MATVIEW)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * Check if it is a temp table (presumably, of some other backend's).
@@ -2025,8 +2029,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2091,6 +2094,7 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		/*
 		 * We cannot safely process other backends' temp tables, so skip 'em.
@@ -2099,6 +2103,9 @@ do_autovacuum(void)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * fetch reloptions -- if this toast table does not have them, try the
@@ -2118,8 +2125,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2916,8 +2922,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	PgStat_StatTabEntry *tabentry;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index bc8c43b96aa..89bf0cbed56 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,37 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +197,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +215,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -213,20 +251,23 @@ pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
 
+	locator = rel->rd_locator;
+
 	/* Store the data in the table's hash table entry. */
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel), false);
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
+											false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -285,9 +326,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -325,9 +366,25 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -468,7 +525,16 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
+}
+
+PgStat_StatTabEntry *
+pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator)
+{
+	return (PgStat_StatTabEntry *) pgstat_fetch_entry(
+													  PGSTAT_KIND_RELATION,
+													  locator.dbOid,
+													  RelFileLocatorToPgStatObjid(locator));
 }
 
 /*
@@ -476,12 +542,14 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
+
+	return pgstat_fetch_stat_tabentry_by_locator(locator);
 }
 
 /*
@@ -503,14 +571,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -706,8 +777,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -750,7 +825,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -785,8 +860,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -920,17 +995,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1009,3 +1088,82 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index 50ea9e8fb83..3f59739de7c 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -23,13 +23,13 @@
 #include "common/ip.h"
 #include "funcapi.h"
 #include "miscadmin.h"
-#include "pgstat.h"
 #include "postmaster/bgworker.h"
 #include "replication/logicallauncher.h"
 #include "storage/proc.h"
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 #include "utils/wait_event.h"
 
@@ -1962,9 +1962,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2318,5 +2323,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/catalog/pg_tablespace.dat b/src/include/catalog/pg_tablespace.dat
index c4cde415219..73ed046be31 100644
--- a/src/include/catalog/pg_tablespace.dat
+++ b/src/include/catalog/pg_tablespace.dat
@@ -10,6 +10,10 @@
 #
 #----------------------------------------------------------------------
 
+/*
+ * When adding a new one, ensure it does not conflict with
+ * PSEUDO_PARTITION_TABLE_SPCOID.
+ */
 [
 
 { oid => '1663', oid_symbol => 'DEFAULTTABLESPACE_OID',
diff --git a/src/include/catalog/pg_tablespace.h b/src/include/catalog/pg_tablespace.h
index 3bd4a74f003..e5e818eb0c5 100644
--- a/src/include/catalog/pg_tablespace.h
+++ b/src/include/catalog/pg_tablespace.h
@@ -21,6 +21,14 @@
 #include "catalog/genbki.h"
 #include "catalog/pg_tablespace_d.h"	/* IWYU pragma: export */
 
+/*
+ * Reserved tablespace OID for partitioned table pseudo locators.
+ * This is not an actual tablespace, just a reserved value to distinguish
+ * partitioned table statistics from regular table statistics. Ensures it does
+ * not conflict with the ones in pg_tablespace.dat.
+ */
+#define PSEUDO_PARTITION_TABLE_SPCOID 1665
+
 /* ----------------
  *		pg_tablespace definition.  cpp turns this into
  *		typedef struct FormData_pg_tablespace
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index 216b93492ba..689c624f373 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -15,6 +15,7 @@
 #include "portability/instr_time.h"
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -38,6 +39,12 @@ typedef struct RelationData *Relation;
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -178,11 +185,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -738,8 +745,8 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index 9b8fbae00ed..42f29496551 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -765,6 +765,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index cdc427dbc78..4d00087dc6f 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--sdA5zuaL94ts0rne
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v12-0002-handle-relation-statistics-correctly-during-rewr.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread

* [PATCH v13 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

This is possible as, since b14e9ce7d55c, the objoid is now uint64 and spcOid
and relNumber are 32 bits.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/postmaster/autovacuum.c          |  17 +-
 src/backend/utils/activity/pgstat_relation.c | 236 ++++++++++++++++---
 src/backend/utils/adt/pgstatfuncs.c          |  21 +-
 src/include/catalog/pg_tablespace.dat        |   4 +
 src/include/catalog/pg_tablespace.h          |   8 +
 src/include/pgstat.h                         |  15 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 8 files changed, 271 insertions(+), 71 deletions(-)
   6.1% src/backend/postmaster/
  61.7% src/backend/utils/activity/
   4.9% src/backend/utils/adt/
   3.2% src/include/catalog/
   5.6% src/include/
  18.1% src/test/recovery/t/

diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index 219673db930..4bef7ab7543 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -1992,12 +1992,16 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		if (classForm->relkind != RELKIND_RELATION &&
 			classForm->relkind != RELKIND_MATVIEW)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * Check if it is a temp table (presumably, of some other backend's).
@@ -2026,8 +2030,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2092,6 +2095,7 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		/*
 		 * We cannot safely process other backends' temp tables, so skip 'em.
@@ -2100,6 +2104,9 @@ do_autovacuum(void)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * fetch reloptions -- if this toast table does not have them, try the
@@ -2119,8 +2126,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2917,8 +2923,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	PgStat_StatTabEntry *tabentry;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index bc8c43b96aa..89bf0cbed56 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,37 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +197,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +215,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -213,20 +251,23 @@ pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
 
+	locator = rel->rd_locator;
+
 	/* Store the data in the table's hash table entry. */
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel), false);
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
+											false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -285,9 +326,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -325,9 +366,25 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -468,7 +525,16 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
+}
+
+PgStat_StatTabEntry *
+pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator)
+{
+	return (PgStat_StatTabEntry *) pgstat_fetch_entry(
+													  PGSTAT_KIND_RELATION,
+													  locator.dbOid,
+													  RelFileLocatorToPgStatObjid(locator));
 }
 
 /*
@@ -476,12 +542,14 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
+
+	return pgstat_fetch_stat_tabentry_by_locator(locator);
 }
 
 /*
@@ -503,14 +571,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -706,8 +777,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -750,7 +825,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -785,8 +860,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -920,17 +995,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1009,3 +1088,82 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index 5f907335990..106d1e577a7 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -30,6 +30,7 @@
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 #include "utils/tuplestore.h"
 #include "utils/wait_event.h"
@@ -1965,9 +1966,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2324,5 +2330,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/catalog/pg_tablespace.dat b/src/include/catalog/pg_tablespace.dat
index c4cde415219..73ed046be31 100644
--- a/src/include/catalog/pg_tablespace.dat
+++ b/src/include/catalog/pg_tablespace.dat
@@ -10,6 +10,10 @@
 #
 #----------------------------------------------------------------------
 
+/*
+ * When adding a new one, ensure it does not conflict with
+ * PSEUDO_PARTITION_TABLE_SPCOID.
+ */
 [
 
 { oid => '1663', oid_symbol => 'DEFAULTTABLESPACE_OID',
diff --git a/src/include/catalog/pg_tablespace.h b/src/include/catalog/pg_tablespace.h
index 3bd4a74f003..e5e818eb0c5 100644
--- a/src/include/catalog/pg_tablespace.h
+++ b/src/include/catalog/pg_tablespace.h
@@ -21,6 +21,14 @@
 #include "catalog/genbki.h"
 #include "catalog/pg_tablespace_d.h"	/* IWYU pragma: export */
 
+/*
+ * Reserved tablespace OID for partitioned table pseudo locators.
+ * This is not an actual tablespace, just a reserved value to distinguish
+ * partitioned table statistics from regular table statistics. Ensures it does
+ * not conflict with the ones in pg_tablespace.dat.
+ */
+#define PSEUDO_PARTITION_TABLE_SPCOID 1665
+
 /* ----------------
  *		pg_tablespace definition.  cpp turns this into
  *		typedef struct FormData_pg_tablespace
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index 216b93492ba..689c624f373 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -15,6 +15,7 @@
 #include "portability/instr_time.h"
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -38,6 +39,12 @@ typedef struct RelationData *Relation;
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -178,11 +185,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -738,8 +745,8 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index 9b8fbae00ed..42f29496551 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -765,6 +765,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index cdc427dbc78..4d00087dc6f 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--USlF2Vr25us5UpvY
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v13-0002-handle-relation-statistics-correctly-during-rewr.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread

* [PATCH v10 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

This is possible as, since b14e9ce7d55c, the objoid is now uint64 and spcOid
and relNumber are 32 bits.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/postmaster/autovacuum.c          |  17 +-
 src/backend/utils/activity/pgstat_relation.c | 236 ++++++++++++++++---
 src/backend/utils/adt/pgstatfuncs.c          |  22 +-
 src/include/catalog/pg_tablespace.dat        |   4 +
 src/include/catalog/pg_tablespace.h          |   8 +
 src/include/pgstat.h                         |  15 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 8 files changed, 271 insertions(+), 72 deletions(-)
   6.1% src/backend/postmaster/
  61.6% src/backend/utils/activity/
   5.1% src/backend/utils/adt/
   3.2% src/include/catalog/
   5.6% src/include/
  18.1% src/test/recovery/t/

diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index 1bd3924e35e..a11174b25ad 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -2014,12 +2014,16 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		if (classForm->relkind != RELKIND_RELATION &&
 			classForm->relkind != RELKIND_MATVIEW)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * Check if it is a temp table (presumably, of some other backend's).
@@ -2048,8 +2052,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2114,6 +2117,7 @@ do_autovacuum(void)
 		bool		dovacuum;
 		bool		doanalyze;
 		bool		wraparound;
+		RelFileLocator locator;
 
 		/*
 		 * We cannot safely process other backends' temp tables, so skip 'em.
@@ -2122,6 +2126,9 @@ do_autovacuum(void)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * fetch reloptions -- if this toast table does not have them, try the
@@ -2141,8 +2148,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2939,8 +2945,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	PgStat_StatTabEntry *tabentry;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index 55a10c299db..a267e93f8be 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,37 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +197,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +215,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -213,20 +251,23 @@ pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
 
+	locator = rel->rd_locator;
+
 	/* Store the data in the table's hash table entry. */
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel), false);
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
+											false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -285,9 +326,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -325,9 +366,25 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -468,7 +525,16 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
+}
+
+PgStat_StatTabEntry *
+pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator)
+{
+	return (PgStat_StatTabEntry *) pgstat_fetch_entry(
+													  PGSTAT_KIND_RELATION,
+													  locator.dbOid,
+													  RelFileLocatorToPgStatObjid(locator));
 }
 
 /*
@@ -476,12 +542,14 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
+
+	return pgstat_fetch_stat_tabentry_by_locator(locator);
 }
 
 /*
@@ -503,14 +571,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -706,8 +777,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -750,7 +825,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -785,8 +860,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -920,17 +995,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1009,3 +1088,82 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index ef6fffe60b9..60ffb1679ec 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -23,13 +23,13 @@
 #include "common/ip.h"
 #include "funcapi.h"
 #include "miscadmin.h"
-#include "pgstat.h"
 #include "postmaster/bgworker.h"
 #include "replication/logicallauncher.h"
 #include "storage/proc.h"
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 
 #define UINT32_ACCESS_ONCE(var)		 ((uint32)(*((volatile uint32 *)&(var))))
@@ -1949,9 +1949,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2305,5 +2310,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/catalog/pg_tablespace.dat b/src/include/catalog/pg_tablespace.dat
index 1302a3d75cd..9430970fffd 100644
--- a/src/include/catalog/pg_tablespace.dat
+++ b/src/include/catalog/pg_tablespace.dat
@@ -10,6 +10,10 @@
 #
 #----------------------------------------------------------------------
 
+/*
+ * When adding a new one, ensure it does not conflict with
+ * PSEUDO_PARTITION_TABLE_SPCOID.
+ */
 [
 
 { oid => '1663', oid_symbol => 'DEFAULTTABLESPACE_OID',
diff --git a/src/include/catalog/pg_tablespace.h b/src/include/catalog/pg_tablespace.h
index 7816d779d8c..0e2d8051d69 100644
--- a/src/include/catalog/pg_tablespace.h
+++ b/src/include/catalog/pg_tablespace.h
@@ -21,6 +21,14 @@
 #include "catalog/genbki.h"
 #include "catalog/pg_tablespace_d.h"	/* IWYU pragma: export */
 
+/*
+ * Reserved tablespace OID for partitioned table pseudo locators.
+ * This is not an actual tablespace, just a reserved value to distinguish
+ * partitioned table statistics from regular table statistics. Ensures it does
+ * not conflict with the ones in pg_tablespace.dat.
+ */
+#define PSEUDO_PARTITION_TABLE_SPCOID 1665
+
 /* ----------------
  *		pg_tablespace definition.  cpp turns this into
  *		typedef struct FormData_pg_tablespace
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index 6714363144a..3102f86aa24 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -17,6 +17,7 @@
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
 #include "replication/worker_internal.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -35,6 +36,12 @@
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -175,11 +182,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -735,8 +742,8 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index 5c1ce4d3d6a..7b24928b00d 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -764,6 +764,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index 021e2bf361f..3a9c05eaf10 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--xJsrX25x3tm7Tn81
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v10-0002-handle-relation-statistics-correctly-during-rewr.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread

* [PATCH v15 1/2] Key PGSTAT_KIND_RELATION by relfile locator
@ 2025-10-01 09:45 Bertrand Drouvot <[email protected]>
  0 siblings, 0 replies; 291+ messages in thread

From: Bertrand Drouvot @ 2025-10-01 09:45 UTC (permalink / raw)

This patch changes the key used for the PGSTAT_KIND_RELATION statistic kind.
Instead of the relation oid, it now relies on:

- dboid (linked to RelFileLocator's dbOid)
- objoid which is the result of a new macro (namely RelFileLocatorToPgStatObjid())
that computes an objoid based on the RelFileLocator's spcOid and the
RelFileLocator's relNumber.

This is possible as, since b14e9ce7d55c, the objoid is now uint64 and spcOid
and relNumber are 32 bits.

That will allow us to add new stats (add writes counters) and ensure that some
counters (n_dead_tup and friends) are replicated.

The patch introduces pgstat_reloid_to_relfilelocator() to 1) avoid calling
RelationIdGetRelation() to get the relfilelocator based on the relation oid
and 2) handle the partitioned table case.

Please note that:

- when running pg_stat_have_stats('relation',...) we now need to be connected
to the database that hosts the relation. As pg_stat_have_stats() is not
documented publicly, then the changes done in 029_stats_restart.pl look
enough.

- this patch does not handle rewrites so some tests are failing. It's only
intent is to ease the review and should not be pushed without being
merged with the following patch that handles the rewrites.

- it can be used to test that stats are incremented correctly and that we're
able to retrieve them as long as rewrites are not involved.
---
 src/backend/postmaster/autovacuum.c          |  18 +-
 src/backend/utils/activity/pgstat_relation.c | 236 ++++++++++++++++---
 src/backend/utils/adt/pgstatfuncs.c          |  21 +-
 src/include/catalog/pg_tablespace.dat        |   4 +
 src/include/catalog/pg_tablespace.h          |   8 +
 src/include/pgstat.h                         |  15 +-
 src/include/utils/pgstat_internal.h          |   1 +
 src/test/recovery/t/029_stats_restart.pl     |  40 ++--
 8 files changed, 272 insertions(+), 71 deletions(-)
   6.2% src/backend/postmaster/
  61.7% src/backend/utils/activity/
   4.9% src/backend/utils/adt/
   3.2% src/include/catalog/
   5.6% src/include/
  18.1% src/test/recovery/t/

diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index d695f1de4bd..0a028b18d66 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -2035,6 +2035,7 @@ do_autovacuum(void)
 		bool		doanalyze;
 		bool		wraparound;
 		AutoVacuumScores scores;
+		RelFileLocator locator;
 
 		if (classForm->relkind != RELKIND_RELATION &&
 			classForm->relkind != RELKIND_MATVIEW)
@@ -2042,6 +2043,10 @@ do_autovacuum(void)
 
 		relid = classForm->oid;
 
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
+
 		/*
 		 * Check if it is a temp table (presumably, of some other backend's).
 		 * We cannot safely process other backends' temp tables.
@@ -2069,8 +2074,7 @@ do_autovacuum(void)
 
 		/* Fetch reloptions and the pgstat entry for this table */
 		relopts = extract_autovac_opts(tuple, pg_class_desc);
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		/* Check if it needs vacuum or analyze */
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
@@ -2143,6 +2147,7 @@ do_autovacuum(void)
 		bool		doanalyze;
 		bool		wraparound;
 		AutoVacuumScores scores;
+		RelFileLocator locator;
 
 		/*
 		 * We cannot safely process other backends' temp tables, so skip 'em.
@@ -2151,6 +2156,9 @@ do_autovacuum(void)
 			continue;
 
 		relid = classForm->oid;
+		locator.dbOid = classForm->relisshared ? InvalidOid : MyDatabaseId;
+		locator.spcOid = classForm->reltablespace;
+		locator.relNumber = classForm->relfilenode;
 
 		/*
 		 * fetch reloptions -- if this toast table does not have them, try the
@@ -2170,8 +2178,7 @@ do_autovacuum(void)
 		}
 
 		/* Fetch the pgstat entry for this table */
-		tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-												  relid);
+		tabentry = pgstat_fetch_stat_tabentry_by_locator(locator);
 
 		relation_needs_vacanalyze(relid, relopts, classForm, tabentry,
 								  effective_multixact_freeze_max_age,
@@ -2988,8 +2995,7 @@ recheck_relation_needs_vacanalyze(Oid relid,
 	AutoVacuumScores scores;
 
 	/* fetch the pgstat table entry */
-	tabentry = pgstat_fetch_stat_tabentry_ext(classForm->relisshared,
-											  relid);
+	tabentry = pgstat_fetch_stat_tabentry_ext(relid);
 
 	relation_needs_vacanalyze(relid, avopts, classForm, tabentry,
 							  effective_multixact_freeze_max_age,
diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c
index bc8c43b96aa..89bf0cbed56 100644
--- a/src/backend/utils/activity/pgstat_relation.c
+++ b/src/backend/utils/activity/pgstat_relation.c
@@ -17,12 +17,17 @@
 
 #include "postgres.h"
 
+#include "access/htup_details.h"
 #include "access/twophase_rmgr.h"
 #include "access/xact.h"
 #include "catalog/catalog.h"
+#include "catalog/pg_tablespace.h"
+#include "storage/lmgr.h"
 #include "utils/memutils.h"
 #include "utils/pgstat_internal.h"
 #include "utils/rel.h"
+#include "utils/relmapper.h"
+#include "utils/syscache.h"
 #include "utils/timestamp.h"
 
 
@@ -36,13 +41,12 @@ typedef struct TwoPhasePgStatRecord
 	PgStat_Counter inserted_pre_truncdrop;
 	PgStat_Counter updated_pre_truncdrop;
 	PgStat_Counter deleted_pre_truncdrop;
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	RelFileLocator locator;		/* table's rd_locator */
 	bool		truncdropped;	/* was the relation truncated/dropped? */
 } TwoPhasePgStatRecord;
 
 
-static PgStat_TableStatus *pgstat_prep_relation_pending(Oid rel_id, bool isshared);
+static PgStat_TableStatus *pgstat_prep_relation_pending(RelFileLocator locator);
 static void add_tabstat_xact_level(PgStat_TableStatus *pgstat_info, int nest_level);
 static void ensure_tabstat_xact_level(PgStat_TableStatus *pgstat_info);
 static void save_truncdrop_counters(PgStat_TableXactStatus *trans, bool is_drop);
@@ -60,8 +64,7 @@ pgstat_copy_relation_stats(Relation dst, Relation src)
 	PgStatShared_Relation *dstshstats;
 	PgStat_EntryRef *dst_ref;
 
-	srcstats = pgstat_fetch_stat_tabentry_ext(src->rd_rel->relisshared,
-											  RelationGetRelid(src));
+	srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src));
 	if (!srcstats)
 		return;
 
@@ -94,8 +97,10 @@ pgstat_init_relation(Relation rel)
 
 	/*
 	 * We only count stats for relations with storage and partitioned tables
+	 * and we don't count stats generated during a rewrite.
 	 */
-	if (!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE)
+	if ((!RELKIND_HAS_STORAGE(relkind) && relkind != RELKIND_PARTITIONED_TABLE) ||
+		OidIsValid(rel->rd_rel->relrewrite))
 	{
 		rel->pgstat_enabled = false;
 		rel->pgstat_info = NULL;
@@ -130,12 +135,37 @@ pgstat_init_relation(Relation rel)
 void
 pgstat_assoc_relation(Relation rel)
 {
+	RelFileLocator locator;
+
 	Assert(rel->pgstat_enabled);
 	Assert(rel->pgstat_info == NULL);
 
+	/*
+	 * Don't associate stats for relations without storage and non partitioned
+	 * tables.
+	 */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind) &&
+		rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		return;
+
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
+
 	/* Else find or make the PgStat_TableStatus entry, and update link */
-	rel->pgstat_info = pgstat_prep_relation_pending(RelationGetRelid(rel),
-													rel->rd_rel->relisshared);
+	rel->pgstat_info = pgstat_prep_relation_pending(locator);
 
 	/* don't allow link a stats to multiple relcache entries */
 	Assert(rel->pgstat_info->relation == NULL);
@@ -167,9 +197,13 @@ pgstat_unlink_relation(Relation rel)
 void
 pgstat_create_relation(Relation rel)
 {
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_create_transactional(PGSTAT_KIND_RELATION,
-								rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-								RelationGetRelid(rel));
+								rel->rd_locator.dbOid,
+								RelFileLocatorToPgStatObjid(rel->rd_locator));
 }
 
 /*
@@ -181,9 +215,13 @@ pgstat_drop_relation(Relation rel)
 	int			nest_level = GetCurrentTransactionNestLevel();
 	PgStat_TableStatus *pgstat_info;
 
+	/* don't track stats for relations without storage */
+	if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
+		return;
+
 	pgstat_drop_transactional(PGSTAT_KIND_RELATION,
-							  rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId,
-							  RelationGetRelid(rel));
+							  rel->rd_locator.dbOid,
+							  RelFileLocatorToPgStatObjid(rel->rd_locator));
 
 	if (!pgstat_should_count_relation(rel))
 		return;
@@ -213,20 +251,23 @@ pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
 
+	locator = rel->rd_locator;
+
 	/* Store the data in the table's hash table entry. */
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel), false);
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
+											false);
 
 	shtabentry = (PgStatShared_Relation *) entry_ref->shared_stats;
 	tabentry = &shtabentry->stats;
@@ -285,9 +326,9 @@ pgstat_report_analyze(Relation rel,
 	PgStat_EntryRef *entry_ref;
 	PgStatShared_Relation *shtabentry;
 	PgStat_StatTabEntry *tabentry;
-	Oid			dboid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
 	TimestampTz ts;
 	PgStat_Counter elapsedtime;
+	RelFileLocator locator;
 
 	if (!pgstat_track_counts)
 		return;
@@ -325,9 +366,25 @@ pgstat_report_analyze(Relation rel,
 	ts = GetCurrentTimestamp();
 	elapsedtime = TimestampDifferenceMilliseconds(starttime, ts);
 
+	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
+		locator = rel->rd_locator;
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator.dbOid = (rel->rd_rel->relisshared ? InvalidOid : MyDatabaseId);
+		locator.spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator.relNumber = rel->rd_id;
+	}
 	/* block acquiring lock for the same reason as pgstat_report_autovac() */
-	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, dboid,
-											RelationGetRelid(rel),
+	entry_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION,
+											locator.dbOid,
+											RelFileLocatorToPgStatObjid(locator),
 											false);
 	/* can't get dropped while accessed */
 	Assert(entry_ref != NULL && entry_ref->shared_stats != NULL);
@@ -468,7 +525,16 @@ pgstat_update_heap_dead_tuples(Relation rel, int delta)
 PgStat_StatTabEntry *
 pgstat_fetch_stat_tabentry(Oid relid)
 {
-	return pgstat_fetch_stat_tabentry_ext(IsSharedRelation(relid), relid);
+	return pgstat_fetch_stat_tabentry_ext(relid);
+}
+
+PgStat_StatTabEntry *
+pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator)
+{
+	return (PgStat_StatTabEntry *) pgstat_fetch_entry(
+													  PGSTAT_KIND_RELATION,
+													  locator.dbOid,
+													  RelFileLocatorToPgStatObjid(locator));
 }
 
 /*
@@ -476,12 +542,14 @@ pgstat_fetch_stat_tabentry(Oid relid)
  * whether the to-be-accessed table is a shared relation or not.
  */
 PgStat_StatTabEntry *
-pgstat_fetch_stat_tabentry_ext(bool shared, Oid reloid)
+pgstat_fetch_stat_tabentry_ext(Oid reloid)
 {
-	Oid			dboid = (shared ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	return (PgStat_StatTabEntry *)
-		pgstat_fetch_entry(PGSTAT_KIND_RELATION, dboid, reloid);
+	if (!pgstat_reloid_to_relfilelocator(reloid, &locator))
+		return NULL;
+
+	return pgstat_fetch_stat_tabentry_by_locator(locator);
 }
 
 /*
@@ -503,14 +571,17 @@ find_tabstat_entry(Oid rel_id)
 	PgStat_TableXactStatus *trans;
 	PgStat_TableStatus *tabentry = NULL;
 	PgStat_TableStatus *tablestatus = NULL;
+	RelFileLocator locator;
+
+	if (!pgstat_reloid_to_relfilelocator(rel_id, &locator))
+		return NULL;
+
+	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION,
+										   locator.dbOid,
+										   RelFileLocatorToPgStatObjid(locator));
 
-	entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, MyDatabaseId, rel_id);
 	if (!entry_ref)
-	{
-		entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, InvalidOid, rel_id);
-		if (!entry_ref)
-			return tablestatus;
-	}
+		return tablestatus;
 
 	tabentry = (PgStat_TableStatus *) entry_ref->pending;
 	tablestatus = palloc_object(PgStat_TableStatus);
@@ -706,8 +777,12 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state)
 		record.inserted_pre_truncdrop = trans->inserted_pre_truncdrop;
 		record.updated_pre_truncdrop = trans->updated_pre_truncdrop;
 		record.deleted_pre_truncdrop = trans->deleted_pre_truncdrop;
-		record.id = tabstat->id;
-		record.shared = tabstat->shared;
+
+		if (tabstat->relation != NULL)
+			record.locator = tabstat->relation->rd_locator;
+		else
+			record.locator = tabstat->locator;
+
 		record.truncdropped = trans->truncdropped;
 
 		RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0,
@@ -750,7 +825,7 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info,
 	PgStat_TableStatus *pgstat_info;
 
 	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, commit case */
 	pgstat_info->counts.tuples_inserted += rec->tuples_inserted;
@@ -785,8 +860,8 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 	TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata;
 	PgStat_TableStatus *pgstat_info;
 
-	/* Find or create a tabstat entry for the rel */
-	pgstat_info = pgstat_prep_relation_pending(rec->id, rec->shared);
+	/* Find or create a tabstat entry for the target locator */
+	pgstat_info = pgstat_prep_relation_pending(rec->locator);
 
 	/* Same math as in AtEOXact_PgStat, abort case */
 	if (rec->truncdropped)
@@ -920,17 +995,21 @@ pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts)
  * initialized if not exists.
  */
 static PgStat_TableStatus *
-pgstat_prep_relation_pending(Oid rel_id, bool isshared)
+pgstat_prep_relation_pending(RelFileLocator locator)
 {
 	PgStat_EntryRef *entry_ref;
 	PgStat_TableStatus *pending;
+	uint64		objid;
+
+	objid = RelFileLocatorToPgStatObjid(locator);
 
 	entry_ref = pgstat_prep_pending_entry(PGSTAT_KIND_RELATION,
-										  isshared ? InvalidOid : MyDatabaseId,
-										  rel_id, NULL);
+										  locator.dbOid,
+										  objid, NULL);
+
 	pending = entry_ref->pending;
-	pending->id = rel_id;
-	pending->shared = isshared;
+	pending->id = objid;
+	pending->locator = locator;
 
 	return pending;
 }
@@ -1009,3 +1088,82 @@ restore_truncdrop_counters(PgStat_TableXactStatus *trans)
 		trans->tuples_deleted = trans->deleted_pre_truncdrop;
 	}
 }
+
+/*
+ * Convert a relation OID to its corresponding RelFileLocator for statistics
+ * tracking purposes.
+ *
+ * Returns true on success, false if the relation doesn't need statistics
+ * tracking.
+ *
+ * For partitioned tables, constructs a synthetic locator using the relation
+ * OID as relNumber, since they don't have storage.
+ */
+bool
+pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator)
+{
+	HeapTuple	tuple;
+	Form_pg_class relform;
+	bool		result = true;
+
+	/* get the relation's tuple from pg_class */
+	tuple = SearchSysCache1(RELOID, ObjectIdGetDatum(reloid));
+
+	if (!HeapTupleIsValid(tuple))
+		return false;
+
+	relform = (Form_pg_class) GETSTRUCT(tuple);
+
+	/* skip relations without storage and non partitioned tables */
+	if (!RELKIND_HAS_STORAGE(relform->relkind) &&
+		relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		ReleaseSysCache(tuple);
+		return false;
+	}
+
+	if (relform->relkind != RELKIND_PARTITIONED_TABLE)
+	{
+		/* build the RelFileLocator */
+		locator->relNumber = relform->relfilenode;
+		locator->spcOid = relform->reltablespace;
+
+		/* handle default tablespace */
+		if (!OidIsValid(locator->spcOid))
+			locator->spcOid = MyDatabaseTableSpace;
+
+		/* handle dbOid for global vs local relations */
+		if (locator->spcOid == GLOBALTABLESPACE_OID)
+			locator->dbOid = InvalidOid;
+		else
+			locator->dbOid = MyDatabaseId;
+
+		/* handle mapped relations */
+		if (!RelFileNumberIsValid(locator->relNumber))
+		{
+			locator->relNumber = RelationMapOidToFilenumber(reloid,
+															relform->relisshared);
+			if (!RelFileNumberIsValid(locator->relNumber))
+			{
+				ReleaseSysCache(tuple);
+				return false;
+			}
+		}
+	}
+	else
+	{
+		/*
+		 * Partitioned tables don't have storage, so construct a synthetic
+		 * locator for statistics tracking. Use a reserved pseudo tablespace
+		 * OID that cannot conflict with real tablespaces, and the relation
+		 * OID as relNumber. This ensures no collision with regular relations
+		 * even after OID wraparound.
+		 */
+		locator->dbOid = (relform->relisshared ? InvalidOid : MyDatabaseId);
+		locator->spcOid = PSEUDO_PARTITION_TABLE_SPCOID;
+		locator->relNumber = relform->oid;
+	}
+
+	ReleaseSysCache(tuple);
+	return result;
+}
diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c
index 9185a8e6b83..db882339f1e 100644
--- a/src/backend/utils/adt/pgstatfuncs.c
+++ b/src/backend/utils/adt/pgstatfuncs.c
@@ -30,6 +30,7 @@
 #include "storage/procarray.h"
 #include "utils/acl.h"
 #include "utils/builtins.h"
+#include "utils/pgstat_internal.h"
 #include "utils/timestamp.h"
 #include "utils/tuplestore.h"
 #include "utils/wait_event.h"
@@ -2004,9 +2005,14 @@ Datum
 pg_stat_reset_single_table_counters(PG_FUNCTION_ARGS)
 {
 	Oid			taboid = PG_GETARG_OID(0);
-	Oid			dboid = (IsSharedRelation(taboid) ? InvalidOid : MyDatabaseId);
+	RelFileLocator locator;
 
-	pgstat_reset(PGSTAT_KIND_RELATION, dboid, taboid);
+	/* Get the stats locator from the relation OID */
+	if (!pgstat_reloid_to_relfilelocator(taboid, &locator))
+		PG_RETURN_VOID();
+
+	pgstat_reset(PGSTAT_KIND_RELATION, locator.dbOid,
+				 RelFileLocatorToPgStatObjid(locator));
 
 	PG_RETURN_VOID();
 }
@@ -2363,5 +2369,16 @@ pg_stat_have_stats(PG_FUNCTION_ARGS)
 	uint64		objid = PG_GETARG_INT64(2);
 	PgStat_Kind kind = pgstat_get_kind_from_str(stats_type);
 
+	/* Convert relation OID to relfilenode objid */
+	if (kind == PGSTAT_KIND_RELATION)
+	{
+		RelFileLocator locator;
+
+		if (!pgstat_reloid_to_relfilelocator(objid, &locator))
+			PG_RETURN_BOOL(false);
+
+		objid = RelFileLocatorToPgStatObjid(locator);
+	}
+
 	PG_RETURN_BOOL(pgstat_have_entry(kind, dboid, objid));
 }
diff --git a/src/include/catalog/pg_tablespace.dat b/src/include/catalog/pg_tablespace.dat
index c4cde415219..73ed046be31 100644
--- a/src/include/catalog/pg_tablespace.dat
+++ b/src/include/catalog/pg_tablespace.dat
@@ -10,6 +10,10 @@
 #
 #----------------------------------------------------------------------
 
+/*
+ * When adding a new one, ensure it does not conflict with
+ * PSEUDO_PARTITION_TABLE_SPCOID.
+ */
 [
 
 { oid => '1663', oid_symbol => 'DEFAULTTABLESPACE_OID',
diff --git a/src/include/catalog/pg_tablespace.h b/src/include/catalog/pg_tablespace.h
index 3bd4a74f003..e5e818eb0c5 100644
--- a/src/include/catalog/pg_tablespace.h
+++ b/src/include/catalog/pg_tablespace.h
@@ -21,6 +21,14 @@
 #include "catalog/genbki.h"
 #include "catalog/pg_tablespace_d.h"	/* IWYU pragma: export */
 
+/*
+ * Reserved tablespace OID for partitioned table pseudo locators.
+ * This is not an actual tablespace, just a reserved value to distinguish
+ * partitioned table statistics from regular table statistics. Ensures it does
+ * not conflict with the ones in pg_tablespace.dat.
+ */
+#define PSEUDO_PARTITION_TABLE_SPCOID 1665
+
 /* ----------------
  *		pg_tablespace definition.  cpp turns this into
  *		typedef struct FormData_pg_tablespace
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index 8e3549c3752..ef740aff26e 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -16,6 +16,7 @@
 #include "postmaster/pgarch.h"	/* for MAX_XFN_CHARS */
 #include "replication/conflict.h"
 #include "storage/locktag.h"
+#include "storage/relfilelocator.h"
 #include "utils/backend_progress.h" /* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/backend_status.h"	/* for backward compatibility */	/* IWYU pragma: export */
 #include "utils/pgstat_kind.h"
@@ -39,6 +40,12 @@ typedef struct RelationData *Relation;
 /* Default directory to store temporary statistics data in */
 #define PG_STAT_TMP_DIR		"pg_stat_tmp"
 
+/*
+ * Build a pgstat key Objid based on a RelFileLocator.
+ */
+#define RelFileLocatorToPgStatObjid(locator) \
+	(((uint64) (locator).spcOid << 32) | (locator).relNumber)
+
 /* Values for track_functions GUC variable --- order is significant! */
 typedef enum TrackFunctionsLevel
 {
@@ -179,11 +186,11 @@ typedef struct PgStat_TableCounts
  */
 typedef struct PgStat_TableStatus
 {
-	Oid			id;				/* table's OID */
-	bool		shared;			/* is it a shared catalog? */
+	uint64		id;				/* hash of relfilelocator for stats key */
 	struct PgStat_TableXactStatus *trans;	/* lowest subxact's counts */
 	PgStat_TableCounts counts;	/* event counts to be sent */
 	Relation	relation;		/* rel that is using this entry */
+	RelFileLocator locator;		/* table's relfilelocator */
 } PgStat_TableStatus;
 
 /* ----------
@@ -766,8 +773,8 @@ extern void pgstat_twophase_postabort(FullTransactionId fxid, uint16 info,
 									  void *recdata, uint32 len);
 
 extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry(Oid relid);
-extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(bool shared,
-														   Oid reloid);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_by_locator(RelFileLocator locator);
+extern PgStat_StatTabEntry *pgstat_fetch_stat_tabentry_ext(Oid reloid);
 extern PgStat_TableStatus *find_tabstat_entry(Oid rel_id);
 
 
diff --git a/src/include/utils/pgstat_internal.h b/src/include/utils/pgstat_internal.h
index eed4c6b359c..1090f049058 100644
--- a/src/include/utils/pgstat_internal.h
+++ b/src/include/utils/pgstat_internal.h
@@ -786,6 +786,7 @@ extern void PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state);
 extern bool pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait);
 extern void pgstat_relation_delete_pending_cb(PgStat_EntryRef *entry_ref);
 extern void pgstat_relation_reset_timestamp_cb(PgStatShared_Common *header, TimestampTz ts);
+extern bool pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator);
 
 
 /*
diff --git a/src/test/recovery/t/029_stats_restart.pl b/src/test/recovery/t/029_stats_restart.pl
index cdc427dbc78..4d00087dc6f 100644
--- a/src/test/recovery/t/029_stats_restart.pl
+++ b/src/test/recovery/t/029_stats_restart.pl
@@ -55,10 +55,10 @@ trigger_funcrel_stat();
 
 # verify stats objects exist
 $sect = "initial";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -79,10 +79,10 @@ copy($og_stats, $statsfile) or die "Copy failed: $!";
 $node->start;
 
 $sect = "copy";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 $node->stop('immediate');
@@ -96,10 +96,10 @@ $node->start;
 
 # stats should have been discarded
 $sect = "post immediate";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 # get rid of backup statsfile
@@ -110,10 +110,10 @@ unlink $statsfile or die "cannot unlink $statsfile $!";
 trigger_funcrel_stat();
 
 $sect = "post immediate, new";
-is(have_stats('database', $dboid, 0), 't', "$sect: db stats do exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 't', "$sect: db stats do exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	't', "$sect: function stats do exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	't', "$sect: relation stats do exist");
 
 # regular shutdown
@@ -129,10 +129,10 @@ $node->start;
 
 # no stats present due to invalid stats file
 $sect = "invalid_overwrite";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -145,10 +145,10 @@ append_file($og_stats, "XYZ");
 $node->start;
 
 $sect = "invalid_append";
-is(have_stats('database', $dboid, 0), 'f', "$sect: db stats do not exist");
-is(have_stats('function', $dboid, $funcoid),
+is(have_stats($connect_db, 'database', $dboid, 0), 'f', "$sect: db stats do not exist");
+is(have_stats($db_under_test, 'function', $dboid, $funcoid),
 	'f', "$sect: function stats do not exist");
-is(have_stats('relation', $dboid, $tableoid),
+is(have_stats($db_under_test, 'relation', $dboid, $tableoid),
 	'f', "$sect: relation stats do not exist");
 
 
@@ -307,9 +307,9 @@ sub trigger_funcrel_stat
 
 sub have_stats
 {
-	my ($kind, $dboid, $objid) = @_;
+	my ($db, $kind, $dboid, $objid) = @_;
 
-	return $node->safe_psql($connect_db,
+	return $node->safe_psql($db,
 		"SELECT pg_stat_have_stats('$kind', $dboid, $objid)");
 }
 
-- 
2.34.1


--Og8T0Lf2JVsuKOzn
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v15-0002-handle-relation-statistics-correctly-during-rewr.patch"



^ permalink  raw  reply  [nested|flat] 291+ messages in thread