From: Tom Lane <tgl@sss.pgh.pa.us>
To: Nathan Bossart <nathandbossart@gmail.com>
cc: pgsql-hackers@postgresql.org
Subject: Re: remove check hooks for GUCs that contribute to MaxBackends
In-reply-to: <ZnMr2k-Nk5vj7T7H@nathan>
References: <ZnMr2k-Nk5vj7T7H@nathan>
Comments: In-reply-to Nathan Bossart <nathandbossart@gmail.com>
	message dated "Wed, 19 Jun 2024 14:04:58 -0500"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <147691.1718824176.1@sss.pgh.pa.us>
Date: Wed, 19 Jun 2024 15:09:36 -0400
Message-ID: <147692.1718824176@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/147692.1718824176%40sss.pgh.pa.us>
Precedence: bulk

Nathan Bossart <nathandbossart@gmail.com> writes:
> While working on an idea from another thread [0], I noticed that each of
> max_connections, max_worker_process, max_autovacuum_workers, and
> max_wal_senders have a check hook that verifies the sum of those GUCs does
> not exceed a certain value.  Then, in InitializeMaxBackends(), we do the
> same check once more.  Not only do the check hooks seem redundant, but I
> think they might sometimes be inaccurate since some values might not yet be
> initialized.

Yeah, these per-variable checks are inherently bogus.  If we can get
of them and make the net user experience actually better, that's a
win-win.

It seems easier to do for these because they can't change after server
start, so there can be one well-defined time to apply the consistency
check.  IIRC, we have some similar issues in other hooks for variables
that aren't PGC_POSTMASTER, so it's harder to see how we might get rid
of their cross-checks.  That doesn't make them less bogus though.

			regards, tom lane