Message-Id: <471F390A.EE98.0025.0@wicourts.gov>
Date: Wed, 24 Oct 2007 12:22:35 -0500
From: "Kevin Grittner" <Kevin.Grittner@wicourts.gov>
To: "Honza Novak" <kacerr@developers.zlutazimnice.cz>,
	<pgsql-performance@postgresql.org>
Subject: Re: multiple apaches against single postgres
	database
References: <471F3752.8070205@developers.zlutazimnice.cz>
In-Reply-To: <471F3752.8070205@developers.zlutazimnice.cz>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

>>> On Wed, Oct 24, 2007 at  7:15 AM, in message
<471F3752.8070205@developers.zlutazimnice.cz>, Honza Novak
<kacerr@developers.zlutazimnice.cz> wrote:=20
=20
> But sometimes (i don't know exactly for what reason) some queries =
gets=20
> stuck (mostly they are inserts or updates, but realy simple) and=20
> postgres is unable to answer in time
=20
In addition to the points made by others, there is a chance that a
contributing factor is the tendency of PostgreSQL (prior to the
upcoming 8.3 release) to hold onto dirty pages for as long as
possible and throw them all at the disk drives in at checkpoint
time.  In some such cases the advice from previous emails may not
be enough -- you may have to use very aggressive background writer
settings, a smaller shared buffers setting, and/or reduce or
eliminate the OS write delays.
=20
If you find this to be your problem, you may want to be an early
adopter of the 8.3 release, once it is out.
=20
-Kevin
=20