On Wed, Jun 21, 2023 at 09:26:09AM -0700, Jeff Davis wrote: > What I meant is that if you do: > > CREATE TABLE p(i INT, j INT) PARTITION BY RANGE (i); > CREATE TABLE p0 PARTITION OF p FOR VALUES FROM (00) TO (10); > CREATE TABLE p1 PARTITION OF p FOR VALUES FROM (10) TO (20); > CREATE INDEX p_idx ON p (i); > CREATE INDEX special_idx ON p0 (j); > GRANT MAINTAIN ON p TO foo; > \c - foo > REINDEX TABLE p; > > That would reindex p0_i_idx and p1_i_idx, but skip special_idx. That > might be too confusing, but feels a bit more consistent permissions- > wise.
FWIW, the current behavior to reindex special_idx in this case feels more natural to me, as the user requests a REINDEX at table-level, not at index-level. This would mean to me that all the indexes of all the partitions should be rebuilt on, not just the partitioned indexes that are defined in the partitioned table requested for rebuild. -- Michael
signature.asc
Description: PGP signature
