Movatterモバイル変換

[0]ホーム
URL:

Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

 Sign up
Appearance settings

Commit9eefba1

Browse files
committed
Delay lock acquisition for partitions until we route a tuple to them.
Instead of locking all partitions to which we might route a tuple atexecutor startup, just lock them as we use them. In some cases such apartition might get locked at executor startup anyway because itappears in the query's range table for some other reason, but in othercases this is a bit savings.This changes the order in which partitions are locked in some cases,which might conceivably create deadlock hazards that don't existtoday, but per discussion, it seems like such cases should be rareenough that we can neglect them in favor of improving performance.David Rowley, reviewed and tested by Tomas Vondra, Sho Kato, JohnNaylor, Tom Lane, and me.Discussion:http://postgr.es/m/CAKJS1f-=FnMqmQP6qitkD+xEddxw22ySLP-0xFk3JAqUX2yfMw@mail.gmail.com
1 parent7aa00d2 commit9eefba1

File tree

1 file changed

+16
-19
lines changed

1 file changed

+16
-19
lines changed

‎src/backend/executor/execPartition.c

Lines changed: 16 additions & 19 deletions
Original file line numberDiff line numberDiff line change
@@ -191,9 +191,6 @@ static void find_matching_subplans_recurse(PartitionPruningData *prunedata,
191191
* tuple routing for partitioned tables, encapsulates it in
192192
* PartitionTupleRouting, and returns it.
193193
*
194-
* Note that all the relations in the partition tree are locked using the
195-
* RowExclusiveLock mode upon return from this function.
196-
*
197194
* Callers must use the returned PartitionTupleRouting during calls to
198195
* ExecFindPartition(). The actual ResultRelInfo for a partition is only
199196
* allocated when the partition is found for the first time.
@@ -208,9 +205,6 @@ ExecSetupPartitionTupleRouting(ModifyTableState *mtstate, Relation rel)
208205
PartitionTupleRouting*proute;
209206
ModifyTable*node=mtstate ? (ModifyTable*)mtstate->ps.plan :NULL;
210207

211-
/* Lock all the partitions. */
212-
(void)find_all_inheritors(RelationGetRelid(rel),RowExclusiveLock,NULL);
213-
214208
/*
215209
* Here we attempt to expend as little effort as possible in setting up
216210
* the PartitionTupleRouting. Each partition's ResultRelInfo is built on
@@ -487,8 +481,9 @@ ExecHashSubPlanResultRelsByOid(ModifyTableState *mtstate,
487481

488482
/*
489483
* ExecInitPartitionInfo
490-
*Initialize ResultRelInfo and other information for a partition
491-
*and store it in the next empty slot in the proute->partitions array.
484+
*Lock the partition and initialize ResultRelInfo. Also setup other
485+
*information for the partition and store it in the next empty slot in
486+
*the proute->partitions array.
492487
*
493488
* Returns the ResultRelInfo
494489
*/
@@ -510,11 +505,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
510505

511506
oldcxt=MemoryContextSwitchTo(proute->memcxt);
512507

513-
/*
514-
* We locked all the partitions in ExecSetupPartitionTupleRouting
515-
* including the leaf partitions.
516-
*/
517-
partrel=table_open(dispatch->partdesc->oids[partidx],NoLock);
508+
partrel=table_open(dispatch->partdesc->oids[partidx],RowExclusiveLock);
518509

519510
leaf_part_rri=makeNode(ResultRelInfo);
520511
InitResultRelInfo(leaf_part_rri,
@@ -964,11 +955,12 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
964955

965956
/*
966957
* ExecInitPartitionDispatchInfo
967-
*Initialize PartitionDispatch for a partitioned table and store it in
968-
*the next available slot in the proute->partition_dispatch_info array.
969-
*Also, record the index into this array in the parent_pd->indexes[]
970-
*array in the partidx element so that we can properly retrieve the
971-
*newly created PartitionDispatch later.
958+
*Lock the partitioned table (if not locked already) and initialize
959+
*PartitionDispatch for a partitioned table and store it in the next
960+
*available slot in the proute->partition_dispatch_info array. Also,
961+
*record the index into this array in the parent_pd->indexes[] array in
962+
*the partidx element so that we can properly retrieve the newly created
963+
*PartitionDispatch later.
972964
*/
973965
staticPartitionDispatch
974966
ExecInitPartitionDispatchInfo(PartitionTupleRouting*proute,Oidpartoid,
@@ -982,8 +974,13 @@ ExecInitPartitionDispatchInfo(PartitionTupleRouting *proute, Oid partoid,
982974

983975
oldcxt=MemoryContextSwitchTo(proute->memcxt);
984976

977+
/*
978+
* Only sub-partitioned tables need to be locked here. The root
979+
* partitioned table will already have been locked as it's referenced in
980+
* the query's rtable.
981+
*/
985982
if (partoid!=RelationGetRelid(proute->partition_root))
986-
rel=table_open(partoid,NoLock);
983+
rel=table_open(partoid,RowExclusiveLock);
987984
else
988985
rel=proute->partition_root;
989986
partdesc=RelationGetPartitionDesc(rel);

0 commit comments

Comments
 (0)
[8]ページ先頭
©2009-2025 Movatter.jp