Re: autovacuum questions

From: "Scott Marlowe" <scott(dot)marlowe(at)gmail(dot)com>
To: "Alvaro Herrera" <alvherre(at)commandprompt(dot)com>
Cc: "Rafael Domiciano" <rafael(dot)domiciano(at)gmail(dot)com>, isabella(dot)ghiurea(at)nrc-cnrc(dot)gc(dot)ca, pgsql-admin(at)postgresql(dot)org
Subject: Re: autovacuum questions
Date: 2008-10-31 18:31:56
Message-ID: dcc563d10810311131h68980c73gdfc6ada9bdfda12@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-admin

On Fri, Oct 31, 2008 at 10:02 AM, Alvaro Herrera
<alvherre(at)commandprompt(dot)com> wrote:
> Rafael Domiciano escribió:
>
>> The vacuum is not full, to run a full vacuum you need to set a cron job;
>> better to run at the night.
>
> Doing VACUUM FULL on crontab is rarely a good idea. If you find
> yourself in a situation where you need VACUUM FULL, then you've not
> tuned regular vacuum appropriately.

There are some use cases where vacuum full is appropriate. But
they're rare. And they're usually better off being added to whatever
script is doing the thing that causes the database to need vacuum
full. Certain batch processing or data loading processes need vacuum
full and or reindex when finishing.

But the common thought process on vacuum full for most people is "If
vacuum is good, vacuum full must be even better!" which is just wrong.

In response to

Browse pgsql-admin by date

  From Date Subject
Next Message Peter Hinse 2008-10-31 21:27:56 hanging INSERT statements with subselect
Previous Message Alvaro Herrera 2008-10-31 16:02:13 Re: autovacuum questions