Horton

Re: BUG #18256: spgGetCache() attempts to read to read invalid file for index attached to partitioned table

Tom Lane <tgl@sss.pgh.pa.us>

View thread View on postgr.es
From: Tom Lane <tgl@sss.pgh.pa.us>
To: exclusion@gmail.com
Cc: pgsql-bugs@lists.postgresql.org
Date: 2023-12-21T16:04:16Z
Lists: pgsql-bugs 
PG Bug reporting form <noreply@postgresql.org> writes:
> The following query:
> CREATE TABLE t(a int, b text) PARTITION BY RANGE (a);
> CREATE INDEX idx ON t USING spgist(b);
> SELECT * FROM t;

> produces an unexpected error:
> ERROR:  could not open file "pg_tblspc/0/PG_16_202307071/0/0": No such file
> or directory

Thanks for the report.  It looks like we could work around this by
teaching spgGetCache to not try to get the lastUsedPages info for
a partitioned index.  (The rest of what it does is still valid for
partitioned indexes.)  I wonder though if there are any other similar
bugs --- obviously, non-btree partitioned indexes haven't been
tested sufficiently.

> Reproduced on REL_16_STABLE (starting from 3c569049b) .. master.

It should be possible to exhibit similar problems long before 3c569049b,
I would guess.

			regards, tom lane

Commits

  1. Avoid trying to fetch metapage of an SPGist partitioned index.

  2. Don't try to open visibilitymap when analyzing a foreign table