Fix costing of parallel hash joins.

petere · petere · commit 5b804cc168a5 · 2018-03-06T21:54:37.000-05:00
Commit 1804284 established that single-batch parallel-aware hash joins could create one large shared hash table using the combined work_mem budget of all participants. The costing accidentally assumed that parallel-oblivious hash joins could also do that. The documentation for initial_cost_hashjoin() also failed to mention the new argument. Repair. Author: Thomas Munro Reported-By: Antonin Houska Reviewed-By: Antonin Houska Discussion: https://postgr.es/m/12441.1513935950%40localhost
diff --git a/src/backend/optimizer/path/costsize.c b/src/backend/optimizer/path/costsize.c
@@ -3143,6 +3143,8 @@ cached_scansel(PlannerInfo *root, RestrictInfo *rinfo, PathKey *pathkey)
  * 'outer_path' is the outer input to the join
  * 'inner_path' is the inner input to the join
  * 'extra' contains miscellaneous information about the join
+ * 'parallel_hash' indicates that inner_path is partial and that a shared
+ *		hash table will be built in parallel
  */
 void
 initial_cost_hashjoin(PlannerInfo *root, JoinCostWorkspace *workspace,
diff --git a/src/backend/optimizer/path/joinpath.c b/src/backend/optimizer/path/joinpath.c
@@ -814,7 +814,7 @@ try_partial_hashjoin_path(PlannerInfo *root,
 	 * cost.  Bail out right away if it looks terrible.
 	 */
 	initial_cost_hashjoin(root, &workspace, jointype, hashclauses,
-						  outer_path, inner_path, extra, true);
+						  outer_path, inner_path, extra, parallel_hash);
 	if (!add_partial_path_precheck(joinrel, workspace.total_cost, NIL))
 		return;
 
