Date: Wed, 22 Feb 2023 13:12:22 -0600
From: Justin Pryzby <pryzby@telsasoft.com>
To: Andres Freund <andres@anarazel.de>
Cc: Heikki Linnakangas <hlinnaka@iki.fi>,
	Duncan Sands <duncan.sands@deepbluecap.com>,
	pgsql-bugs@lists.postgresql.org
Subject: Re: Unlimited memory consumption with long-lived connection
Message-ID: <20230222191221.GE1653@telsasoft.com>
References: <a53cacb0-8835-57d6-31e4-4c5ef196de1a@deepbluecap.com>
 <7484b4bd-a808-ed9e-892c-933f7f24120f@iki.fi>
 <20230222145341.GT1653@telsasoft.com>
 <20230222182811.n6n7bvzc3kd3n6f3@awork3.anarazel.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20230222182811.n6n7bvzc3kd3n6f3@awork3.anarazel.de>
User-Agent: Mutt/1.9.4 (2018-02-28)
Archived-At: 
 <https://www.postgresql.org/message-id/20230222191221.GE1653%40telsasoft.com>
Precedence: bulk

On Wed, Feb 22, 2023 at 10:28:11AM -0800, Andres Freund wrote:
> On 2023-02-22 08:53:41 -0600, Justin Pryzby wrote:

> > There are patches proposed here, which fixed the issue for me.
> > But I've always been suspicious that there may be a 2nd, undiagnosed
> > issue lurking behind this one...
> 
> Any more details?

No.  My suspicion is because the process is killed by OOM roughly every
24h, and the leak didn't seem to be big enough to cause that.

Maybe the leak is nonlinear or not completely deterministic or something
sometimes exacerbates the original issue, or maybe there's only one
issue that's 90% but not 100% understood, or maybe there's a 2nd issue
that's latent once the fix is applied for the known issue.

There's also this report:
https://www.postgresql.org/message-id/1605804350439-0.post%40n3.nabble.com

..which offers no details or reponse, and could easily be a leak in
someting other than JIT, or postgres, or not a leak at all.

-- 
Justin