Adjust tuples estimate for appendrels

In set_append_rel_size(), we currently set rel->tuples to rel->rows
for an appendrel.  Generally, rel->tuples is the raw number of tuples
in the relation and rel->rows is the estimated number of tuples after
the relation's restriction clauses have been applied.  Although an
appendrel itself doesn't directly enforce any quals today, its child
relations may.  Therefore, setting rel->tuples equal to rel->rows for
an appendrel isn't always appropriate.

Doing so can lead to issues in cost estimates in some cases.  For
instance, when estimating the number of distinct values from an
appendrel, we would not be able to adjust the estimate based on the
restriction selectivity.

This patch addresses this by setting an appendrel's tuples to the
total number of tuples accumulated from each live child, which better
aligns with reality.

This is arguably a bug, but nobody has complained about that until
now, so no back-patch.

Author: Richard Guo <[email protected]>
Reviewed-by: Tender Wang <[email protected]>
Reviewed-by: Alena Rybakina <[email protected]>
Discussion: https://postgr.es/m/CAMbWs4_TG_+kVn6fjG-5GYzzukrNK57=g9eUo4gsrUG26OFawg@mail.gmail.com

1 files changed, 13 insertions, 6 deletions

diff --git a/src/backend/optimizer/path/allpaths.c b/src/backend/optimizer/path/allpaths.c
index 1115ebeee29..b5bc9b602e2 100644
--- a/src/backend/optimizer/path/allpaths.c
+++ b/src/backend/optimizer/path/allpaths.c
@@ -958,6 +958,7 @@ set_append_rel_size(PlannerInfo *root, RelOptInfo *rel,
 {
 	int			parentRTindex = rti;
 	bool		has_live_children;
+	double		parent_tuples;
 	double		parent_rows;
 	double		parent_size;
 	double	   *parent_attrsizes;
@@ -983,6 +984,15 @@ set_append_rel_size(PlannerInfo *root, RelOptInfo *rel,
 	/*
 	 * Initialize to compute size estimates for whole append relation.
 	 *
+	 * We handle tuples estimates by setting "tuples" to the total number of
+	 * tuples accumulated from each live child, rather than using "rows".
+	 * Although an appendrel itself doesn't directly enforce any quals, its
+	 * child relations may.  Therefore, setting "tuples" equal to "rows" for
+	 * an appendrel isn't always appropriate, and can lead to inaccurate cost
+	 * estimates.  For example, when estimating the number of distinct values
+	 * from an appendrel, we would be unable to adjust the estimate based on
+	 * the restriction selectivity (see estimate_num_groups).
+	 *
 	 * We handle width estimates by weighting the widths of different child
 	 * rels proportionally to their number of rows.  This is sensible because
 	 * the use of width estimates is mainly to compute the total relation
@@ -995,6 +1005,7 @@ set_append_rel_size(PlannerInfo *root, RelOptInfo *rel,
 	 * have zero rows and/or width, if they were excluded by constraints.
 	 */
 	has_live_children = false;
+	parent_tuples = 0;
 	parent_rows = 0;
 	parent_size = 0;
 	nattrs = rel->max_attr - rel->min_attr + 1;
@@ -1161,6 +1172,7 @@ set_append_rel_size(PlannerInfo *root, RelOptInfo *rel,
 		 */
 		Assert(childrel->rows > 0);
 
+		parent_tuples += childrel->tuples;
 		parent_rows += childrel->rows;
 		parent_size += childrel->reltarget->width * childrel->rows;
 
@@ -1207,18 +1219,13 @@ set_append_rel_size(PlannerInfo *root, RelOptInfo *rel,
 		int			i;
 
 		Assert(parent_rows > 0);
+		rel->tuples = parent_tuples;
 		rel->rows = parent_rows;
 		rel->reltarget->width = rint(parent_size / parent_rows);
 		for (i = 0; i < nattrs; i++)
 			rel->attr_widths[i] = rint(parent_attrsizes[i] / parent_rows);
 
 		/*
-		 * Set "raw tuples" count equal to "rows" for the appendrel; needed
-		 * because some places assume rel->tuples is valid for any baserel.
-		 */
-		rel->tuples = parent_rows;
-
-		/*
 		 * Note that we leave rel->pages as zero; this is important to avoid
 		 * double-counting the appendrel tree in total_table_pages.
 		 */