Re: pg_avd

From: "Matthew T(dot) O'Connor" <matthew(at)zeut(dot)net>
To: Olleg Samoylov <olleg(at)jane(dot)telecom(dot)mipt(dot)ru>
Cc: Patches <pgsql-patches(at)postgresql(dot)org>
Subject: Re: pg_avd
Date: 2003-02-18 18:07:34
Message-ID: 1045591654.12295.4.camel@zeutrh80
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-patches

On Tue, 2003-02-18 at 09:04, Olleg Samoylov wrote:
> Hi all,
>
> On one hand pg_autovacuum must give opportunity to launch it without sleep
> loop, for instance for launch it at night by cron, then it vacuumed only
> tables which needed it.

pg_autovacuum does not preclude you from doing normal vacuuming from
cron.

> On other, if pg_autovacuum running as daemon, IMHO better launch it from
> postmaster, in the same manner, as launched statistic collector.

The problem is that it's not a backend process, it's a libpq based
client app. I'm not sure it's a good idea to have the postmaster launch
client external applications, I can't explain why, just seems that way
to me. Any one else have any thoughts?

> Must pg_autovacuum also check transaction ID (XID) numbers to prevent
> transaction ID wraparound? (PostgreSQL v.7.3.1 docs 8.2.3. Preventing
> transaction ID wraparound failures)

Yes, it must, this is one of the "missing features" that I will complete
if people like the overall pg_autovacuum concept.

> And must it reindex in the same manner as contrib/reindexdb? (PostgreSQL
> v.7.3.1 docs 8.3. Routine Reindexing)

Have never thought of it as a reindexing tool. I think that is outside
the scope of pg_autovacuum since reindexing is not a simple thing, not
always wanted, and has concurrency issues (I think).

In response to

  • Re: pg_avd at 2003-02-18 14:04:19 from Olleg Samoylov

Browse pgsql-patches by date

  From Date Subject
Next Message Matthew T. O'Connor 2003-02-18 18:38:33 Re: pg_avd
Previous Message Bruce Momjian 2003-02-18 17:36:35 Re: FAQ addition: deleteing all but one unique row