Re: POC: make mxidoff 64 bits

On Wed, 18 Dec 2024 at 13:21, Heikki Linnakangas <hlinnaka(at)iki(dot)fi> wrote:

>
> Attached is some more cleanup on top of patch set v9, removing more dead
> stuff related to wraparound. I also removed the oldestOffsetKnown
> variable and related code. It was needed to deal with clusters upgraded
> from buggy 9.3 and 9.4 era versions, but now that pg_upgrade will
> rewrite the SLRUs, it's no longer needed.
>
Yep, multixact.c looks correct to me. As for "XXX could use
SimpleLruTruncate()", yes, for sure.
Actually, xl_multixact_truncate.startTruncMemb is also no longer needed, is
it?

> Does the pg_upgrade code work though, if you have that buggy situation
> where oldestOffsetKnown == false ?
>
> >
> > if (!TransactionIdIsValid(*xactptr))
> > {
> > /* Corner case 3: we must be looking at unused
> slot zero */
> > Assert(offset == 0);
> > continue;
> > }
>
> After upgrade, this corner case 3 would *not* happen on offset == 0. So
> looks like we're still missing test coverage for this upgrade corner case.
>
Am I understanding correctly that you want to have a test corresponding to
the buggy 9.3 and 9.4 era versions?
Do you think we could imitate this scenario on a current master branch like
that:
1) generate a couple of offsets segments for the first table;
2) generate more segments for a second table;
3) drop first table;
4) stop pg cluster;
5) remove pg_multixact/offsets/0000
6) upgrade?

PFA, v10-0016-TEST-try-to-replicate-buggy-oldest-offset.patch
This test will fail now, for an obvious reason, but is this case a relevant
one?

--
Best regards,
Maxim Orlov.