Re: Injection points: some tools to wait and wake

Michael Paquier <michael@paquier.xyz>

From: Michael Paquier <michael@paquier.xyz>
To: Bertrand Drouvot <bertranddrouvot.pg@gmail.com>
Cc: "Andrey M. Borodin" <x4mmm@yandex-team.ru>, Postgres hackers <pgsql-hackers@lists.postgresql.org>, Ashutosh Bapat <ashutosh.bapat.oss@gmail.com>
Date: 2024-03-04T01:44:34Z
Lists: pgsql-hackers
On Fri, Mar 01, 2024 at 06:52:45AM +0000, Bertrand Drouvot wrote:
> +	if (defined($backend_type))
> +	{
> +		$backend_type = qq('$backend_type');
> +		$die_message = "the backend type $backend_type";
> +	}
> +	else
> +	{
> +		$backend_type = 'backend_type';
> +		$die_message = 'one backend';
> +
> +	}
> +
> +	$self->poll_query_until(
> +		'postgres', qq[
> +		SELECT count(*) > 0 FROM pg_stat_activity
> +		WHERE backend_type = $backend_type AND wait_event = '$injection_name'
> +	])
> +	  or die
> +	  qq(timed out waiting for $die_message to wait for the injection point '$injection_name');

I was looking at that, and found v3 to be an overkill.  First, I think
that we should encourage callers to pass down a backend_type.  Perhaps
I am wrong to assume so, but that's based on my catalog of tests
waiting in my stack.

A second thing is that this is entirely unrelated to injection points,
because a test may want to wait for a given wait_event on a
backend_type without using the module injection_points.  At the end, I 
have renamed the routine to wait_for_event(), tweaked a bit its
internals, and the result looked fine so I have applied it while
updating 041_checkpoint_at_promote.pl to use it.  Its internals could
always be expanded more depending on the demand for it.
--
Michael

Commits

  1. Add PostgreSQL::Test::Cluster::wait_for_event()

  2. Add regression test for restart points during promotion

  3. injection_points: Add wait and wakeup of processes