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

Commitedbcbe2

Browse files
author
Etsuro Fujita
committed
postgres_fdw: Fix cost estimation for aggregate pushdown.
In commit7012b13, which added support for aggregate pushdown inpostgres_fdw, the expense of evaluating the final scan/join targetcomputed by make_group_input_target() was not accounted for at all incosting aggregate pushdown paths with local statistics. The right fixfor this would be to have a separate upper stage to adjust the finalscan/join relation (see comments for apply_scanjoin_target_to_paths());but for now, fix by adding the tlist eval cost when costing aggregatepushdown paths with local statistics.Apply this to HEAD only to avoid destabilizing existing plan choices.Author: Etsuro FujitaReviewed-By: Antonin HouskaDiscussion:https://postgr.es/m/5C66A056.60007%40lab.ntt.co.jp
1 parent47a338c commitedbcbe2

File tree

2 files changed

+17
-3
lines changed

2 files changed

+17
-3
lines changed

‎contrib/postgres_fdw/postgres_fdw.c

Lines changed: 13 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -2842,12 +2842,18 @@ estimate_path_cost_size(PlannerInfo *root,
28422842
}
28432843
elseif (IS_UPPER_REL(foreignrel))
28442844
{
2845+
RelOptInfo*outerrel=fpinfo->outerrel;
28452846
PgFdwRelationInfo*ofpinfo;
28462847
AggClauseCostsaggcosts;
28472848
doubleinput_rows;
28482849
intnumGroupCols;
28492850
doublenumGroups=1;
28502851

2852+
/* The upper relation should have its outer relation set */
2853+
Assert(outerrel);
2854+
/* and that outer relation should have its reltarget set */
2855+
Assert(outerrel->reltarget);
2856+
28512857
/*
28522858
* This cost model is mixture of costing done for sorted and
28532859
* hashed aggregates in cost_agg(). We are not sure which
@@ -2856,7 +2862,7 @@ estimate_path_cost_size(PlannerInfo *root,
28562862
* and all finalization and run cost are added in total_cost.
28572863
*/
28582864

2859-
ofpinfo= (PgFdwRelationInfo*)fpinfo->outerrel->fdw_private;
2865+
ofpinfo= (PgFdwRelationInfo*)outerrel->fdw_private;
28602866

28612867
/* Get rows and width from input rel */
28622868
input_rows=ofpinfo->rows;
@@ -2909,23 +2915,27 @@ estimate_path_cost_size(PlannerInfo *root,
29092915

29102916
/*-----
29112917
* Startup cost includes:
2912-
* 1. Startup cost for underneath input relation
2918+
* 1. Startup cost for underneath input relation, adjusted for
2919+
* tlist replacement by apply_scanjoin_target_to_paths()
29132920
* 2. Cost of performing aggregation, per cost_agg()
29142921
*-----
29152922
*/
29162923
startup_cost=ofpinfo->rel_startup_cost;
2924+
startup_cost+=outerrel->reltarget->cost.startup;
29172925
startup_cost+=aggcosts.transCost.startup;
29182926
startup_cost+=aggcosts.transCost.per_tuple*input_rows;
29192927
startup_cost+=aggcosts.finalCost.startup;
29202928
startup_cost+= (cpu_operator_cost*numGroupCols)*input_rows;
29212929

29222930
/*-----
29232931
* Run time cost includes:
2924-
* 1. Run time cost of underneath input relation
2932+
* 1. Run time cost of underneath input relation, adjusted for
2933+
* tlist replacement by apply_scanjoin_target_to_paths()
29252934
* 2. Run time cost of performing aggregation, per cost_agg()
29262935
*-----
29272936
*/
29282937
run_cost=ofpinfo->rel_total_cost-ofpinfo->rel_startup_cost;
2938+
run_cost+=outerrel->reltarget->cost.per_tuple*input_rows;
29292939
run_cost+=aggcosts.finalCost.per_tuple*numGroups;
29302940
run_cost+=cpu_tuple_cost*numGroups;
29312941

‎src/backend/optimizer/plan/planner.c

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -7113,6 +7113,10 @@ apply_scanjoin_target_to_paths(PlannerInfo *root,
71137113
* confused in createplan.c if they don't agree. We must do this now so
71147114
* that any append paths made in the next part will use the correct
71157115
* pathtarget (cf. create_append_path).
7116+
*
7117+
* Note that this is also necessary if GetForeignUpperPaths() gets called
7118+
* on the final scan/join relation or on any of its children, since the
7119+
* FDW might look at the rel's target to create ForeignPaths.
71167120
*/
71177121
rel->reltarget=llast_node(PathTarget,scanjoin_targets);
71187122

0 commit comments

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