Subject: Re: Memory usage of auto-vacuum
From: Guillaume Lelarge <guillaume@lelarge.info>
To: Gael Le Mignot <gael@pilotsystems.net>
Cc: pgsql-performance@postgresql.org
Date: Sat, 09 Jul 2011 11:06:16 +0200
In-Reply-To: <plop87hb6vn803.fsf@aoskar.kilobug.org>
References: <plop87tyavncer.fsf@aoskar.kilobug.org>
	<1310200383.2101.8.camel@laptop>
	<plop87liw7n8t0.fsf@aoskar.kilobug.org>
	<1310201594.2101.12.camel@laptop>
	<plop87hb6vn803.fsf@aoskar.kilobug.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
Message-ID: <1310202377.2101.15.camel@laptop>
Mime-Version: 1.0

On Sat, 2011-07-09 at 11:00 +0200, Gael Le Mignot wrote:
> Hello Guillaume!
> 
> Sat, 09 Jul 2011 10:53:14 +0200, you wrote: 
> 
>  > I don't quite understand how you can get up to 1GB used by your process.
>  > According to your configuration, and unless I'm wrong, it shouldn't take
>  > more than 40MB. Perhaps a bit more, but not 1GB. So, how did you find
>  > this number?
> 
> Looking at  "top" we  saw the postgres  process growing and  growing and
> then shrinking  back, and doing  a "select * from  pg_stat_activity;" in
> parallel of the growing we found only the "vacuum analyze" query running. 
> 

There is not only one postgres process. So you first need to be sure
that it's the one that executes the autovacuum.

> But maybe  we drawn the conclusion  too quickly, I'll  try disabling the
> auto vacuum to see if we really get rid of the problem doing it.
> 

Disabling the autovacuum is usually a bad idea. You'll have to execute
VACUUM/ANALYZE via cron, which could get hard to configure.

BTW, what's your PostgreSQL release? I assume at least 8.3 since you're
using FTS?


-- 
Guillaume
  http://blog.guillaume.lelarge.info
  http://www.dalibo.com