Re: Single transaction in the tablesync worker?

On Mon, Jan 25, 2021 at 4:48 PM Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
>
> On Mon, Jan 25, 2021 at 8:03 AM Peter Smith <smithpb2250(at)gmail(dot)com> wrote:
> >
> > Hi Amit.
> >
> > PSA the v19 patch for the Tablesync Solution1.
> >
>
> I see one race condition in this patch where we try to drop the origin
> via apply process and DropSubscription. I think it can lead to the
> error "cache lookup failed for replication origin with oid %u". The
> same problem can happen via exposed API pg_replication_origin_drop but
> probably because this is not used concurrently so nobody faced this
> issue. I think for the matter of this patch we can try to suppress
> such an error either via try..catch, or by adding missing_ok argument
> to replorigin_drop API, or we can just add to comments that such a
> race exists.

OK. This has been isolated to a common function called from 3 places.
The potential race ERROR is suppressed by TRY/CATCH.
Please see code of latest patch [v20]

> Additionally, we should try to start a new thread for the
> existence of this problem in pg_replication_origin_drop. What do you
> think?

OK. It is on my TODO list..

----
[v20] https://www.postgresql.org/message-id/CAHut%2BPuNwSujoL_dwa%3DTtozJ_vF%3DCnJxjgQTCmNBkazd8J1m-A%40mail.gmail.com

Kind Regards,
Peter Smith.
Fujitsu Australia