Re: Question about VACUUM

From: "Kevin Grittner" <Kevin(dot)Grittner(at)wicourts(dot)gov>
To: "Josh Berkus" <josh(at)agliodbs(dot)com>, <pgsql-performance(at)postgresql(dot)org>
Subject: Re: Question about VACUUM
Date: 2011-12-07 15:14:11
Message-ID: 4EDF2E630200002500043986@gw.wicourts.gov
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

Josh Berkus <josh(at)agliodbs(dot)com> wrote:
> On 12/5/11 1:36 PM, Kevin Grittner wrote:
>> I understand the impulse to run autovacuum less frequently or
>> less aggressively. When we first started running PostgreSQL the
>> default configuration was very cautious.
>
> The default settings are deliberately cautious, as default
> settings should be.

I was talking historically, about the defaults in 8.1:

http://www.postgresql.org/docs/8.1/interactive/runtime-config-autovacuum.html

Those defaults were *over*-cautious to the point that we experienced
serious problems. My point was that many people's first instinct in
that case is to make the setting less aggressive, as I initially did
and the OP has done. The problem is actually solved by making them
*more* aggressive. Current defaults are pretty close to what we
found, through experimentation, worked well for us for most
databases.

> But yes, anyone with a really large/high-traffic database will
> often want to make autovac more aggressive.

I think we're in agreement: current defaults are good for a typical
environment; high-end setups still need to tune to more aggressive
settings. This is an area where incremental changes with monitoring
works well.

-Kevin

In response to

Browse pgsql-performance by date

  From Date Subject
Next Message Christiaan Willemsen 2011-12-07 15:15:52 Partitions and joins lead to index lookups on all partitions
Previous Message Marti Raudsepp 2011-12-07 14:56:53 Re: Intersect/Union X AND/OR