To: "Anibal David Acosta" <aa@devshock.com>
cc: "'Craig Ringer'" <ringerc@ringerc.id.au>, pgsql-performance@postgresql.org
Subject: Re: autovacuum, exclude table 
In-reply-to: <010a01ccb8de$07c0ad20$17420760$@devshock.com> 
References: <00f501ccb8d9$f07dc4d0$d1794e70$@devshock.com>
	<CAD2md3EpKvvEwA_+ZggC0iSO5qCNZ3_eZ25N9Gc7cqLkV3FAPg@mail.gmail.com>
	<010a01ccb8de$07c0ad20$17420760$@devshock.com>
Comments: In-reply-to "Anibal David Acosta" <aa@devshock.com>
	message dated "Mon, 12 Dec 2011 11:55:00 -0300"
Date: Mon, 12 Dec 2011 10:26:10 -0500
Message-ID: <18792.1323703570@sss.pgh.pa.us>
From: Tom Lane <tgl@sss.pgh.pa.us>

"Anibal David Acosta" <aa@devshock.com> writes:
> Tables has insert (in bulk every 3 minutes) and delete one per day (delete records older than XX days)

No updates at all, just inserts and a daily delete?

If so, you're wasting your time even thinking about suppressing
autovacuum, because it won't fire on this table except after the daily
delete, which is exactly when you need it to.

Also, if you suppress autovacuum you also suppress autoanalyze,
which is something that *will* fire after large inserts, and probably
should.  At least, this usage pattern doesn't suggest to me that it's
clearly safe to run without up-to-date stats.

Right offhand, I'm not convinced either that you have a problem, or that
turning off autovacuum would fix it.

			regards, tom lane