Re: AutoVacuum Behaviour Question

From: Alvaro Herrera <alvherre(at)commandprompt(dot)com>
To: Bruce McAlister <bruce(dot)mcalister(at)blueface(dot)ie>
Cc: Martijn van Oosterhout <kleptog(at)svana(dot)org>, pgsql-general(at)postgresql(dot)org
Subject: Re: AutoVacuum Behaviour Question
Date: 2007-06-28 14:00:40
Views: Raw Message | Whole Thread | Download mbox | Resend email
Lists: pgsql-general pgsql-hackers

Bruce McAlister wrote:
> Martijn van Oosterhout wrote:
> > All the values here look OK, except one:
> >
> > On Thu, Jun 28, 2007 at 07:50:36AM +0100, Bruce McAlister wrote:
> >> blueface-crm=# select oid, relfrozenxid from pg_class where relkind in
> >> ('r', 't');
> >> oid | relfrozenxid
> >> ---------+--------------
> >> 2570051 | 2947120794
> >
> > Whatever this table is, the freeze XID isn't getting updated for some
> > reason...


> This looks like a temporary relation,
> temp4295 | 2947120794
> Is there a way we can manually force these to update?

No. Only the session that created the temp table can vacuum it.
Autovacuum skips temp tables. I guess the only thing you can do here is
close that session.

I'm thinking that maybe should make vac_update_datfrozenxid ignore temp
tables. But this doesn't really work, because if we were to truncate
pg_clog there would be tuples on the temp table marked with XIDs that
are nowhere to be found. Maybe we could make some noise about it

This is a problem only in recent releases (8.2) because we started
allowing the max freeze age be configurable.

Alvaro Herrera
PostgreSQL Replication, Consulting, Custom Development, 24x7 support

In response to


Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2007-06-28 14:25:54 Re: Doc update for pg_start_backup
Previous Message Greg Smith 2007-06-28 12:55:50 Re: Bgwriter LRU cleaning: we've been going at this all wrong

Browse pgsql-general by date

  From Date Subject
Next Message Alvaro Herrera 2007-06-28 14:08:06 Re: Execution variability
Previous Message Masaru Sugawara 2007-06-28 13:25:00 Re: Possible bug (or I don't understand how foreign keys should work with partitions)