From: Chris Browne <cbbrowne@acm.org>
Subject: Re: [DOCS] Autovacuum and XID wraparound
Date: Mon, 14 May 2007 16:25:08 -0400
Organization: cbbrowne Computing Inc
Lines: 40
Message-ID: <60sl9z6tln.fsf@dba2.int.libertyrms.com>
References: <20070514011740.GH14860@fetter.org>
	<23394.1179108400@sss.pgh.pa.us> <1179117269.6047.3.camel@goldbach>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
User-Agent: Gnus/5.1007 (Gnus v5.10.7) XEmacs/21.4.20 (linux)
Cancel-Lock: sha1:pkIKf+SSsZulfUOhPoXgjUC9hHU=
To: pgsql-patches@postgresql.org

neilc@samurai.com (Neil Conway) writes:
> On Sun, 2007-13-05 at 22:06 -0400, Tom Lane wrote:
>> This fact is already documented in at least three places; do we really
>> need two more?
>
> I think we need to at least modify the documentation for the autovacuum
> GUC parameter, which currently states only that it "controls whether the
> server should run the autovacuum launcher daemon" -- this is not
> strictly true, and in any case, it isn't the whole story.
>
>> The proposed addition to postgresql.conf seems particularly
>> over-the-top
>
> I agree that this information doesn't really belong in postgresql.conf.

Question... (note: this does not strictly fit into the purview of the
.patches list)

Would the following 'maintenance' regimen be truly safe against XID
wraparound:

 - Most tables are being vacuumed regularly, so that
   pg_class.relfrozenxid is kept "safe."

 - There are some tables that periodically get TRUNCATEd so that, in
   principle, they never need to be vacuumed.

Is it actually true that we'd never need to vacuum those tables
(assuming 8.2+)?  I suppose it would be rather cheap to VACUUM
immediately after the TRUNCATE...

The application is one where we might use partitioning, rolling from
table to table every so often, with the expectation that we'll
TRUNCATE the eldest data often enough that we shouldn't need to VACUUM
any of the partitions.
-- 
let name="cbbrowne" and tld="linuxdatabases.info" in name ^ "@" ^ tld;;
http://linuxdatabases.info/info/linuxxian.html
Why are men like blenders?
You need one, but you're not quite sure why.