Re: pgsql: Allow pgoutput to send logical decoding messages. - Mailing list pgsql-committers

From Amit Kapila
Subject Re: pgsql: Allow pgoutput to send logical decoding messages.
Date
Msg-id CAA4eK1+faTVhxSrVthLxP6N1Smo5cdAarLtOjQwKND+meyMpJQ@mail.gmail.com
Whole thread Raw
In response to Re: pgsql: Allow pgoutput to send logical decoding messages.  (Amit Kapila <amit.kapila16@gmail.com>)
List pgsql-committers
On Tue, Apr 6, 2021 at 2:12 PM Amit Kapila <amit.kapila16@gmail.com> wrote:
>
> On Tue, Apr 6, 2021 at 8:55 AM Amit Kapila <akapila@postgresql.org> wrote:
> >
> > Allow pgoutput to send logical decoding messages.
> >
>
> One of the tests added by this commit is failing:
> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=sidewinder&dt=2021-04-06%2005%3A45%3A20
>
> I think there is a race condition in one of the tests. I'll analyze
> this further.
>

As this is a timing issue so not easy to reproduce but as per my
analysis in the tests, after disabling the subscription, we were not
waiting for the replication connection to drop from the publisher. So
later when the test was trying to use the same slot to fetch the
messages via SQL API, it sometimes gives an error that the replication
slot is active for other PID. I have pushed a fix for this.

-- 
With Regards,
Amit Kapila.



pgsql-committers by date:

Previous
From: Amit Kapila
Date:
Subject: pgsql: Fix the tests added by commit ac4645c015.
Next
From: Peter Eisentraut
Date:
Subject: pgsql: ALTER SUBSCRIPTION ... ADD/DROP PUBLICATION