Re: Additional stats for Relations

From: "Jim C(dot) Nasby" <jim(at)nasby(dot)net>
To: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
Cc: NikhilS <nikkhils(at)gmail(dot)com>, Simon Riggs <simon(at)2ndquadrant(dot)com>, pgsql-hackers(at)postgresql(dot)org
Subject: Re: Additional stats for Relations
Date: 2006-10-19 16:25:35
Message-ID: 20061019162535.GB71084@nasby.net
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Thu, Oct 19, 2006 at 11:47:53AM -0400, Tom Lane wrote:
> "Jim C. Nasby" <jim(at)nasby(dot)net> writes:
> > BTW, if we add these counters we'll be up to 7 stats dealing with vacuum
> > and analyze, and NikhilS has a patch we're finalizing that would add 3
> > more. Right now there's 4 slated to go into pg_stat_* in 8.2, but maybe
> > we should have a separate view for vacuum/FSM statistics?
>
> I've seen no demonstration of a need for *any* of them, actually,
> and am pretty dubious that we want to add so much collection overhead.
> At least not without a major redesign of the stats reporting mechanism.
> If we just drop in another seven counters, we'll create an immediate 50%
> increase in the stats-file I/O volume, even when no vacuuming is
> happening at all.

Yeah, for stuff like vacuuming the current stats system may not make any
sense. Almost anything dealing with vacuum can really just be put into a
table, because it doesn't happen all that often.

My concern is that there's enough useful data to collect about vacuuming
and the FSM that it should probably get it's own set of tables/views,
rather than piggy-backing on pg_stat_*. But that ship has pretty much
sailed, so we're probably stuck with at least the last_* stuff in
pg_stat_* for the immediate future.
--
Jim Nasby jim(at)nasby(dot)net
EnterpriseDB http://enterprisedb.com 512.569.9461 (cell)

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Jim C. Nasby 2006-10-19 16:50:15 Re: [HACKERS] Bug?
Previous Message Tom Lane 2006-10-19 15:47:53 Re: Additional stats for Relations