Re: [COMMITTERS] pgsql: Fix inadequate locking during get_rel_oids().

[Michael Paquier]

On Mon, Mar 05, 2018 at 04:34:09PM -0800, Andres Freund wrote:
> Scratch that, we should be going down the
>       /* If caller supplied OID, there's nothing we need do here. */
>       if (OidIsValid(vrel->oid))
>       {
>               oldcontext = MemoryContextSwitchTo(vac_context);
>               vacrels = lappend(vacrels, vrel);
>               MemoryContextSwitchTo(oldcontext);
>       }
> branch in expand_vacuum_rel() for autovacuum, so this shouldn't
> matter. Sorry for the noise

Yes, I don't see a problem.  However I can understand that it is easy to
be confused on those code paths if you are not used to them and this
area has changed quite a bit the last years.  Perhaps we could add an
Assert(IsAutoVacuumWorkerProcess()) to reduce the confusion?
--
Michael

signature.asc
Description: PGP signature