Re: More logging for autovacuum

From: Decibel! <decibel(at)decibel(dot)org>
To: Andrew Hammond <andrew(dot)george(dot)hammond(at)gmail(dot)com>
Cc: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Peter Eisentraut <peter_e(at)gmx(dot)net>, pgsql-hackers(at)postgresql(dot)org, "Joshua D(dot) Drake" <jd(at)commandprompt(dot)com>, Dimitri Fontaine <dfontaine(at)hi-media(dot)com>, Gregory Stark <stark(at)enterprisedb(dot)com>
Subject: Re: More logging for autovacuum
Date: 2007-08-07 21:30:35
Message-ID: 20070807213035.GQ25704@nasby.net
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Tue, Aug 07, 2007 at 11:59:03AM -0700, Andrew Hammond wrote:
> On 8/7/07, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> >
> > Peter Eisentraut <peter_e(at)gmx(dot)net> writes:
> > > But INFO is not shown by default.
> >
> > INFO is mostly a hack to try to emulate VACUUM VERBOSE's ancient
> > behavior before we redesigned the elog levels. It's intended for
> > controlling messages that should go to a client because the client
> > asked for them, and usually should NOT go to the log. I think it's
> > 100% inappropriate to use it for messages from a background process
> > that has no client.
> >
>
> Traditional log-level semantics aren't very rich. I think that's the real
> source of this problem: "How do I get the info I need to tune auto-vacuum
> without swamping my system with log IO". While the following isn't
> production ready, it seems to have some really good ideas.

Is logging really the answer for that? ISTM it'd be better to provide
statistics info so that you could monitor autovacuum activity with
things like cricket, SNMP, etc.
--
Decibel!, aka Jim Nasby decibel(at)decibel(dot)org
EnterpriseDB http://enterprisedb.com 512.569.9461 (cell)

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Joshua D. Drake 2007-08-07 21:38:09 Re: More logging for autovacuum
Previous Message Tom Lane 2007-08-07 21:03:16 Re: GIT patch