Add empty BRIN ranges during CREATE INDEX

When building BRIN indexes, the brinbuildCallback only advances to the
next page range when seeing a tuple that doesn't belong to the current
one. This means that the index may end up missing ranges at the end of
the table, if those pages do not contain any indexable tuples.

We tend not to have completely empty pages at the end of a relation, but
this also applies to partial indexes, where the tuples may simply not
match the index predicate. This results in inefficient scans using the
affected BRIN index - without the summaries, the page ranges have to be
read and processed, which consumes I/O and possibly also CPU time.

The existing code already added empty ranges for earlier parts of the
table, this commit makes sure we add them for the ranges at the end of
the table too.

Patch by Matthias van de Meent, with review/improvements by me.

Author: Matthias van de Meent
Reviewed-by: Tomas Vondra
Discussion: https://postgr.es/m/CAEze2WiMsPZg%3DxkvSF_jt4%3D69k6K7gz5B8V2wY3gCGZ%2B1BzCbQ%40mail.gmail.com

2 files changed, 35 insertions, 0 deletions

diff --git a/contrib/pageinspect/expected/brin.out b/contrib/pageinspect/expected/brin.out
index 098ddc202f4..3f6e5174bc6 100644
--- a/contrib/pageinspect/expected/brin.out
+++ b/contrib/pageinspect/expected/brin.out
@@ -89,4 +89,22 @@ SELECT brin_revmap_data(decode(repeat('00', :block_size), 'hex'));
  
 (1 row)
 
+-- Test that partial indexes have all pages, including empty ones.
+CREATE TABLE test2 (a int);
+INSERT INTO test2 SELECT i FROM generate_series(1,1000) s(i);
+-- No rows match the index predicate, make sure the index has the right number
+-- of ranges (same as number of page ranges).
+CREATE INDEX ON test2 USING brin (a) WITH (pages_per_range=1) WHERE (a IS NULL);
+ANALYZE test2;
+-- Does the index have one summary of the relation?
+SELECT (COUNT(*) = (SELECT relpages FROM pg_class WHERE relname = 'test2')) AS ranges_do_match
+ FROM generate_series((SELECT (lastrevmappage + 1) FROM brin_metapage_info(get_raw_page('test2_a_idx', 0))),
+                      (SELECT (relpages - 1) FROM pg_class WHERE relname = 'test2_a_idx')) AS pages(p),
+      LATERAL brin_page_items(get_raw_page('test2_a_idx', p), 'test2_a_idx') AS items;
+ ranges_do_match 
+-----------------
+ t
+(1 row)
+
 DROP TABLE test1;
+DROP TABLE test2;
diff --git a/contrib/pageinspect/sql/brin.sql b/contrib/pageinspect/sql/brin.sql
index 96b4645187e..50f260b8e1f 100644
--- a/contrib/pageinspect/sql/brin.sql
+++ b/contrib/pageinspect/sql/brin.sql
@@ -36,4 +36,21 @@ SELECT brin_page_items(decode(repeat('00', :block_size), 'hex'), 'test1_a_idx');
 SELECT brin_metapage_info(decode(repeat('00', :block_size), 'hex'));
 SELECT brin_revmap_data(decode(repeat('00', :block_size), 'hex'));
 
+-- Test that partial indexes have all pages, including empty ones.
+CREATE TABLE test2 (a int);
+INSERT INTO test2 SELECT i FROM generate_series(1,1000) s(i);
+
+-- No rows match the index predicate, make sure the index has the right number
+-- of ranges (same as number of page ranges).
+CREATE INDEX ON test2 USING brin (a) WITH (pages_per_range=1) WHERE (a IS NULL);
+
+ANALYZE test2;
+
+-- Does the index have one summary of the relation?
+SELECT (COUNT(*) = (SELECT relpages FROM pg_class WHERE relname = 'test2')) AS ranges_do_match
+ FROM generate_series((SELECT (lastrevmappage + 1) FROM brin_metapage_info(get_raw_page('test2_a_idx', 0))),
+                      (SELECT (relpages - 1) FROM pg_class WHERE relname = 'test2_a_idx')) AS pages(p),
+      LATERAL brin_page_items(get_raw_page('test2_a_idx', p), 'test2_a_idx') AS items;
+
 DROP TABLE test1;
+DROP TABLE test2;