From: Matthew T. O'Connor <matthew@zeut.net>
To: Gaetano Mendola <mendola@bigfoot.com>
Cc: pgsql-performance@postgresql.org
Subject: Re: is pg_autovacuum so effective ?
Date: Tue, 22 Feb 2005 15:58:01 -0500
Message-ID: <421B9CD9.6090703@zeut.net> (raw)
In-Reply-To: <cvg29i$2l69$1@news.hub.org>
References: <cvg29i$2l69$1@news.hub.org>
Gaetano Mendola wrote:
>pg_class after the vacuum full for that table
>
> relfilenode | relname | relpages | reltuples
>-------------+----------+----------+-------------
> 18376 | messages | 63307 | 1.60644e+06
>
>
>pg_class before the vacuum full for that table
>
> relfilenode | relname | relpages | reltuples
>-------------+----------+----------+-------------
> 18376 | messages | 69472 | 1.60644e+06
>
>
>
>how was possible accumulate 6000 pages wasted on that table?
>
>Between these two calls:
>[2005-02-22 05:25:03 CET] Performing: VACUUM ANALYZE "public"."messages"
>[2005-02-22 15:20:39 CET] Performing: VACUUM ANALYZE "public"."messages"
>
>1768 rows where inserted, and I had 21578 updated for that rows ( each
>row have a counter incremented for each update ) so that table is not
>so heavy updated
>
>I'm running autovacuum with these parameters:
>pg_autovacuum -d 3 -v 300 -V 0.1 -S 0.8 -a 200 -A 0.1 -D
>
>
>shall I run it in a more aggressive way ? May be I'm missing
>something.
>
Well without thinking too much, I would first ask about your FSM
settings? If they aren't big enought that will cause bloat. Try
bumping your FSM settings and then see if you reach steady state.
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: matthew@zeut.net, mendola@bigfoot.com
Subject: Re: is pg_autovacuum so effective ?
In-Reply-To: <421B9CD9.6090703@zeut.net>
* 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