| From: | Andrey Borodin <x4mmm(at)yandex-team(dot)ru> |
|---|---|
| To: | Alexander Lakhin <exclusion(at)gmail(dot)com> |
| Cc: | Peter Geoghegan <pg(at)bowt(dot)ie>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org> |
| Subject: | Re: nbtree backwards scan test coverage |
| Date: | 2026-07-29 11:45:14 |
| Message-ID: | BE794D18-7963-4FB8-A302-F1BA867945BE@yandex-team.ru |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
> On 29 Jul 2026, at 10:00, Alexander Lakhin <exclusion(at)gmail(dot)com> wrote:
>
> Looks like it is not immune to autovacuum.
How about running the module's isolation tests with autovacuum = off
in a temp config? pg_freespacemap does exactly that for the same
reason (VACUUM-dependent output), test_decoding does a softer variant
(autovacuum_naptime = 1d).
Also, as a wild idea, the pages expected to be deleted could be filled
from a rolled-back transaction instead of DELETE+VACUUM of committed rows -
aborted tuples are removable regardless of concurrent snapshots. But I did
not try it with this particular test.
Best regards, Andrey Borodin.
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Daniel Gustafsson | 2026-07-29 11:46:58 | Re: [PATCH] Remove unused include from analyze.c |
| Previous Message | Daniel Gustafsson | 2026-07-29 11:21:34 | Re: [PATCH v1] Fix pg_basebackup long-path test on Windows |