pg.ddx.io  pgsql-admin@postgresql.org mailing list archive  
help / color / mirror / Atom feed
From: Alvaro Herrera <alvherre@commandprompt.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Benjamin Krajmalnik <kraj@illumen.com>
Cc: pgsql-admin@postgresql.org, "Matthew T. O'Connor" <matthew@zeut.net>
Subject: Re: pg_clog questions
Date: Thu, 25 May 2006 16:43:19 -0400
Message-ID: <20060525204319.GL13700@surnet.cl> (raw)
In-Reply-To: <15596.1148588444@sss.pgh.pa.us>
References: <8511B4970E0D124898E973DF496F9B432516E3@stash.stackdump.local>
	<15596.1148588444@sss.pgh.pa.us>

Tom Lane wrote:
> "Benjamin Krajmalnik" <kraj@illumen.com> writes:
> >>> While checking our server, I noticed quite a few files in the pg_clog
> >>> directory.
> >>> Is there a maintenance task which can be run to purge the files which
> >>> are no longer needed?
> 
> >> VACUUM.  If they're not disappearing, you aren't running an adequate
> >> vacuum regime --- missing out some databases, perhaps?
> 
> > I have autovacuum turned on with the default settings.
> 
> Hm.  I believe that autovac only does database-wide vacuums when it
> thinks they're necessary to prevent transaction wraparound failures.
> Which would mean that it'd let pg_clog grow to something on the order
> of half a gig before any truncation would happen.  That's probably
> insufficiently aggressive :-(
> 
> Alvaro, Matthew, any thoughts about improving that?

Hum, IIRC there is a test somewhere to check pg_database.datvacuumxid
and issue a database-wide vacuum if it gets too old, but the test uses
a-little-less-than-2-billion to fire :-(

This is the code:

        /*
         * We look for the database that most urgently needs a database-wide
         * vacuum.  We decide that a database-wide vacuum is needed 100000
         * transactions sooner than vacuum.c's vac_truncate_clog() would
         * decide to start giving warnings.  If any such db is found, we
         * ignore all other dbs.
         *
         * Unlike vacuum.c, we also look at vacuumxid.  This is so that
         * pg_clog can be kept trimmed to a reasonable size.
         */
        freeze_age = (int32) (nextXid - tmp->frozenxid);
        vacuum_age = (int32) (nextXid - tmp->vacuumxid);
        tmp->age = Max(freeze_age, vacuum_age);

        this_whole_db = (tmp->age >
                         (int32) ((MaxTransactionId >> 3) * 3 - 100000));

Maybe it was foolish to use such a large constant in the vacuumxid case
:-(

> It strikes me that Alvaro's work-in-progress on maintaining per-table
> xmin info would allow truncation of clog without actually insisting on
> a database-wide VACUUM, but that's not going to be any help for
> existing releases.

Yeah.  The patch is almost ready BTW and I was about to post it on
Saturday but I got distracted.  I have a problem with sinval messages
:-(  I'll post it later today.

-- 
Alvaro Herrera                                http://www.CommandPrompt.com/
PostgreSQL Replication, Consulting, Custom Development, 24x7 support



view thread (12+ messages)  latest in thread

Message-ID: <20060525204319.GL13700@surnet.cl>
Permalink:  ../20060525204319.GL13700@surnet.cl/
Also on:    postgresql.org/message-id/20060525204319.GL13700@surnet.cl

 · 

reply

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 from your mail client; the link fills in
  the headers below: reply via email
  Some clients, webmail especially, drop In-Reply-To and References from
  a link, and the reply then starts a new thread: use the mbox below.

  To: pgsql-admin@postgresql.org
  Cc: alvherre@commandprompt.com, tgl@sss.pgh.pa.us, kraj@illumen.com, matthew@zeut.net
  Subject: Re: pg_clog questions
  In-Reply-To: <20060525204319.GL13700@surnet.cl>
  References: <8511B4970E0D124898E973DF496F9B432516E3@stash.stackdump.local> <15596.1148588444@sss.pgh.pa.us> <20060525204319.GL13700@surnet.cl>

* 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