Re: VACUUM DELAY - Mailing list pgsql-hackers

From Scott Marlowe
Subject Re: VACUUM DELAY
Date
Msg-id 1092056476.27166.315.camel@localhost.localdomain
Whole thread Raw
In response to VACUUM DELAY  (Gaetano Mendola <mendola@bigfoot.com>)
List pgsql-hackers
On Mon, 2004-08-09 at 05:19, Gaetano Mendola wrote:
> Hi all,
> I have seen the big debat about to have the delay
> off or on by default.
> 
> Why not enable it by default and introduce a new
> parameter to vacuum command itself ? Something like:
> 
> 
> VACUUM .... WITH DELAY 100;
> 
> 
> this will permit to change easilly the delay in the maintainance
> scripts.

The problem, I believe, is that any delay at all results in a VERY slow
vacuum run (like 3 to 5 times slower) and for some people, this will be
such unexpected behaviour they may believe postgresql is broken, or just
want the older, faster vacuum, especially in a development environment. 
Imagine an increase from 1 to 5 minutes on an otherwise duplicate
database from a 7.4 machine.  

I'll personally be running the delay and autovacuum on any machine I'll
be running, and I think once the autovacuum is integrated, it might make
sense to have a vacuum command just toss an entry in a que saying
"vacuum this table next scheduled run" and return immediately with a
NOTICE: vacuum (on tablex) scheduled.





pgsql-hackers by date:

Previous
From: Joerg Hessdoerfer
Date:
Subject: Re: Postgres development model (was Re: CVS comment)
Next
From: Alvaro Herrera Munoz
Date:
Subject: Re: Analyze using savepoints?