database/postgres
1
0
Fork 0
mirror of https://github.com/postgres/postgres.git synced 2025-07-11 10:01:57 +03:00
Code Activity

Repair logic flaw in cost estimator: cost_nestloop() was estimating CPU

Browse Source 
costs using the inner path's parent->rows count as the number of tuples
processed per inner scan iteration.  This is wrong when we are using an
inner indexscan with indexquals based on join clauses, because the rows
count in a Relation node reflects the selectivity of the restriction
clauses for that rel only.  Upshot was that if join clause was very
selective, we'd drastically overestimate the true cost of the join.
Fix is to calculate correct output-rows estimate for an inner indexscan
when the IndexPath node is created and save it in the path node.
Change of path node doesn't require initdb, since path nodes don't
appear in saved rules.
This commit is contained in:
Tom Lane
2000-03-22 22:08:35 +00:00
parent d825e55c13
commit 1d5e7a6f46
11 changed files with 83 additions and 63 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
src/backend/optimizer/path/indxpath.c
View File

@ -9,7 +9,7 @@
*
*
* IDENTIFICATION
* $Header: /cvsroot/pgsql/src/backend/optimizer/path/indxpath.c,v 1.80 2000/02/15 20:49:16 tgl Exp $
* $Header: /cvsroot/pgsql/src/backend/optimizer/path/indxpath.c,v 1.81 2000/03/22 22:08:33 tgl Exp $
*
*-------------------------------------------------------------------------
*/
@ -1454,6 +1454,26 @@ index_innerjoin(Query *root, RelOptInfo *rel, IndexOptInfo *index,
/* joinrelids saves the rels needed on the outer side of the join */
pathnode->joinrelids = lfirst(outerrelids_list);
/*
* We must compute the estimated number of output rows for the
* indexscan. This is less than rel->rows because of the additional
* selectivity of the join clauses. Since clausegroup may contain
* both restriction and join clauses, we have to do a set union to
* get the full set of clauses that must be considered to compute
* the correct selectivity. (We can't just nconc the two lists;
* then we might have some restriction clauses appearing twice,
* which'd mislead restrictlist_selectivity into double-counting
* their selectivity.)
*/
pathnode->rows = rel->tuples *
restrictlist_selectivity(root,
LispUnion(rel->baserestrictinfo,
clausegroup),
lfirsti(rel->relids));
/* Like costsize.c, force estimate to be at least one row */
if (pathnode->rows < 1.0)
pathnode->rows = 1.0;
cost_index(&pathnode->path, root, rel, index, indexquals, true);
path_list = lappend(path_list, pathnode);