Re: Simple postgresql.conf wizard

From: "Robert Haas" <robertmhaas(at)gmail(dot)com>
To: "Greg Smith" <gsmith(at)gregsmith(dot)com>
Cc: "Gregory Stark" <stark(at)enterprisedb(dot)com>, jd(at)commandprompt(dot)com, "Josh Berkus" <josh(at)agliodbs(dot)com>, pgsql-hackers(at)postgresql(dot)org
Subject: Re: Simple postgresql.conf wizard
Date: 2008-12-04 03:03:08
Message-ID: 603c8f070812031903h16f4ff24qd3a8bcdf6d409f20@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

> What fun. I'm beginning to remember why nobody has ever managed to deliver
> a community tool that helps with this configuration task before.

I have to say I really like this tool. It may not be perfect but it's
a lot easier than trying to do this analysis from scratch. And we are
really only arguing about a handful of settings.

It wouldn't take a lot to convince me that checkpoint_segments=3 is
too low. I easily blew through that testing the bulk-insert tuning
patch. I'm curious why wal_buffers is being set to 512 *
checkpoint_segments. Are they related? The default value for
wal_buffers is only 64 kB, which means someone thought you shouldn't
need much space for this at all, but this suggests a setting in the
4-32 MB range, an increase of ~2 orders of magnitude. For all I know
that could be right but it's a big increase.

Regarding the religious war now in progress, I think it would be
awfully good for someone to offer some thoughts on how to figure out
which particular columns on which particular tables need a higher
statistics target. That might allow us to either (a) build a wizard
that helps you find those problems that could perhaps be used
alongside this one or (b) incorporate those same smarts into core.

...Robert

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Bruce Momjian 2008-12-04 03:13:43 Re: Updates of SE-PostgreSQL 8.4devel patches (r1268)
Previous Message Bruce Momjian 2008-12-04 02:52:39 Re: Re: pgsql: Properly unregister OpenSSL callbacks when libpq is done with