Re: Simplify backend terminate and wait logic in postgres_fdw test

On Fri, Apr 9, 2021 at 5:51 AM Michael Paquier <michael(at)paquier(dot)xyz> wrote:
>
> On Thu, Apr 08, 2021 at 06:27:56PM +0530, Bharath Rupireddy wrote:
> > Agree. Please see the attached patch, I removed a fixed waiting time.
> > Instead of relying on pg_stat_activity, pg_sleep and
> > pg_stat_clear_snapshot, now it depends on pg_terminate_backend and
> > pg_wait_for_backend_termination. This way we could reduce the
> > functions that the procedure terminate_backend_and_wait uses and also
> > the new functions pg_terminate_backend and
> > pg_wait_for_backend_termination gets a test coverage.
>
> + EXIT WHEN is_terminated;
> + SELECT * INTO is_terminated FROM pg_wait_for_backend_termination(pid_v, 1000);
> This is still a regression if the termination takes more than 1s,
> no? In such a case terminate_backend_and_wait() would issue a WARNING
> and pollute the regression test output. I can see the point of what
> you are achieving here, and that's a good idea, but from the point of
> view of the buildfarm the WARNING introduced by aaf0432 is a no-go.

I didn't think of the warning cases, my bad. How about using SET
client_min_messages = 'ERROR'; before we call
pg_wait_for_backend_termination? We can only depend on the return
value of pg_wait_for_backend_termination, when true we can exit. This
way the buildfarm will not see warnings. Thoughts?

With Regards,
Bharath Rupireddy.
EnterpriseDB: http://www.enterprisedb.com