Re: Fix assert failure when decoding XLOG_PARAMETER_CHANGE on primary - Mailing list pgsql-hackers

On Mon, Feb 24, 2025 at 7:13 AM Bertrand Drouvot
<bertranddrouvot.pg@gmail.com> wrote:
>
> Hi,
>
> On Mon, Feb 24, 2025 at 05:01:30PM +0530, Amit Kapila wrote:
> > On Wed, Feb 12, 2025 at 5:22 PM Amit Kapila <amit.kapila16@gmail.com> wrote:
> > >
> > > On Wed, Feb 12, 2025 at 1:16 AM Masahiko Sawada <sawada.mshk@gmail.com> wrote:
> > > >
> > > >
> > > > Agreed, I'm fine with leaving InRecovery in this condition. I think
> > > > the point is whether we should add StandbyMode to the condition or
> > > > not. I think if we do that, we would end up with the same error in the
> > > > above scenario I described. So does the following condition make
> > > > sense?
> > > >
> > > >         if (InRecovery &&
> > > >             xlrec.wal_level < WAL_LEVEL_LOGICAL &&
> > > >             wal_level >= WAL_LEVEL_LOGICAL)
> > > >             InvalidateObsoleteReplicationSlots(RS_INVAL_WAL_LEVEL,
> > > >                                                0, InvalidOid,
> > > >                                                InvalidTransactionId);
> > > >
> > >
> > > This will still be true for crash-recovery as the InRecovery flag will
> > > be true for that case as well. I think we should go with your v2 patch
> > > approach for HEAD and back-branches.
> > >
> >
> > Any opinion on how to proceed here?
>
> As far I'm concerned, I did not change my mind since [1] and think the same i.e:
> go with v2 for HEAD and back-branches.

Agreed too. So I'm going to proceed with backpatching the v2 patch to v16.

Regards,

--
Masahiko Sawada
Amazon Web Services: https://aws.amazon.com