From: Scott Marlowe <scott.marlowe@gmail.com>
To: Thomas Lozza <thomas.lozza@nexustelecom.com>
Cc: pgsql-performance@postgresql.org
Subject: Re: Vacuum and FSM page size
Date: Sun, 27 Jan 2008 17:54:12 -0600
Message-ID: <dcc563d10801271554j52c665a4oc851d7fa7f2a5ead@mail.gmail.com> (raw)
In-Reply-To: <F6B04D920899DF47B1E1A4F40415167E3D4370@gaia.nexus-ag.com>
References: <F6B04D920899DF47B1E1A4F40415167E3D4299@gaia.nexus-ag.com>
<dcc563d10801240748m3a2a3309r983a5087ebb17186@mail.gmail.com>
<F6B04D920899DF47B1E1A4F40415167E3D4370@gaia.nexus-ag.com>
On Jan 27, 2008 5:01 PM, Thomas Lozza <thomas.lozza@nexustelecom.com> wrote:
> Thanks for the advice.
> I used the default settings before, thought though that vacuum was a bit
> aggressive, ie, using too many resources. Now its taking very long. So
> will have to find reasonable settings in between I guess.
>
> On the other hand, if I keep the fsm_page number high enough, the system
> should be fine with a low number of vacuum cycles, right. As memory is
> not really scarce (16G, 32 bit PG though) an x million fsm_page entry
> should be ok. Any thoughts on that?
The issue you then run into is bloat, where you have a table or index
that is 90% or so empty space, and performance on selects might
suffer, especially on larger tables or indexes.
What often works best is to let autovacuum handle most of your tables,
then schedule individual tables to be vacuumed by cron, setting the
nap time for vacuum at 20 or 30 milliseconds so they don't chew up all
of your I/O
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-performance@postgresql.org
Cc: scott.marlowe@gmail.com, thomas.lozza@nexustelecom.com
Subject: Re: Vacuum and FSM page size
In-Reply-To: <dcc563d10801271554j52c665a4oc851d7fa7f2a5ead@mail.gmail.com>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox