On Wed, 8 Jul 2026 at 14:50, Fujii Masao <[email protected]> wrote:
>
> On Thu, Jul 2, 2026 at 6:55 PM Amit Kapila <[email protected]> wrote:
> > > How about something like this instead?
> > >
> > >     This function returns a row of NULL values if the specified relation
> > >     OID does not exist, if it is not a sequence, if the current user lacks
> > >     <literal>SELECT</literal> privilege on the sequence, if the sequence
> > >     is another session's temporary sequence, or if it is an unlogged
> > >     sequence on a standby server.
> > >
> >
> > Sounds reasonable. But after this we don't need the next para to say:
> > "It requires <literal>SELECT</literal> privilege on the sequence.".
> > See attached.
>
> Thanks for the patch! I've pushed it.
>
> > > I've updated the patch furthermore. Attached.
> > >
> >
> > LGTM.
>
> Thanks for the review! I've pushed this as well.

There was an issue reported by Noah at [1], the finding-7 reported
fails with an assert. The assertion can be triggered if a sequence is
dropped concurrently while the sequence synchronization worker is
processing it:
TRAP: failed Assert("!isnull"), File: "sequencesync.c", Line: 298, PID: 25085
0   postgres  ExceptionalCondition
1   postgres  get_and_validate_seq_info
2   postgres  copy_sequences
3   postgres  LogicalRepSyncSequences
4   postgres  start_sequence_sync
5   postgres  SequenceSyncWorkerMain

It can be reproduced with the following steps:
Session 1:
BEGIN;
DROP SEQUENCE s1;

Session 2:
ALTER SUBSCRIPTION ... REFRESH SEQUENCES;

Session 1:
COMMIT;

The assertion occurs because the sequence is dropped after the
sequence synchronization worker has collected the sequence metadata
but before it retrieves the sequence data. In this case,
'has_sequence_privilege_id()' calls 'get_rel_relkind()', which can no
longer find the relation in the syscache and therefore returns 'NULL'.
'get_and_validate_seq_info()' currently assumes that the privilege
column is never 'NULL' and triggers the assertion.

I've attached a patch to handle this case by treating a 'NULL'
privilege value as indicating that the sequence no longer exists and
reporting it through the existing "missing sequence on publisher"
path.
Thoughts?

Regards,
Vignesh

Attachment: v1-0001-Fix-sequence-synchronization-for-concurrently-dro.patch
Description: Binary data

Reply via email to