Re: BUG #17485: Records missing from Primary Key index when doing REINDEX INDEX CONCURRENTLY

x4mmm@yandex-team.ru

From: Andrey Borodin <x4mmm@yandex-team.ru>

To: Michael Paquier <michael@paquier.xyz>

Cc: Alvaro Herrera <alvherre@alvh.no-ip.org>, Петър Славов <pet.slavov@gmail.com>, PostgreSQL mailing lists <pgsql-bugs@lists.postgresql.org>

Date: 2022-05-23T13:06:10Z

Lists: pgsql-bugs

Attachments

skip_vacuumed_tuple.diff (application/octet-stream) patch


> 23 мая 2022 г., в 15:49, Andrey Borodin <x4mmm@yandex-team.ru> написал(а):
> 
> I've transformed Peter's test into TAP test that runs ~20 seconds and reliably reproduces problem on my laptop.

I found out one interesting thing: unindexed tuple (that comes from amcheck scan) does not exist in heap page at the moment of check fail.
I've added ReadBuffer() in case of bloom_lacks_element() and ItemIdHasStorage() is false.
I understand that this description is a too vague, so I attached a patch for amcheck relaxing bt_index_check() so the test would pass.

Best regards, Andrey Borodin.

Commits

Revert changes to CONCURRENTLY that "sped up" Xmin advance
- e28bb8851969 15.0 landed
- 042b584c7f7d 14.4 landed
Fix possible HOT corruption when RECENTLY_DEAD changes to DEAD while pruning.
- 18b87b201f73 15.0 cited