Re: Eternal vacuuming.... - Mailing list pgsql-hackers

From Marc G. Fournier
Subject Re: Eternal vacuuming....
Date
Msg-id Pine.BSF.4.21.0005111503340.82196-100000@hub.org
Whole thread Raw
In response to Re: Eternal vacuuming....  (Tim Perdue <tperdue@valinux.com>)
List pgsql-hackers
On Thu, 11 May 2000, Tim Perdue wrote:

> "Marc G. Fournier" wrote:
> > what does a 'vacuum verbose' show for you?  and youa ren't doing a 'vacuum
> > analyze', are you?
> 
> I believe I did 'vacuum analyze'. If info from 'vacuum verbose' would be
> useful to your team, I can try to set up and reproduce this. I would
> have to create a 3-million row table with an index on it, then delete
> 832,000 rows which I did last nite, then try again.

Okay, vacuum analyze is, from my experiences, atrociously slow ... it
*feels* faster, at least, if you do a simple vacuum first, then do the
analyze, but that might be just perception ...

Can you try just a simple 'vacuum verbose' first, without the analyze, and
see if that also takes 12hrs?

Also, what are you running this on?  Memory?  CPU?

Marc G. Fournier                               scrappy@hub.org
Systems Administrator @ hub.org                    
scrappy@{postgresql|isc}.org                       ICQ#7615664



pgsql-hackers by date:

Previous
From: Tim Perdue
Date:
Subject: Re: Eternal vacuuming....
Next
From: The Hermit Hacker
Date:
Subject: Re: Proposal: replace no-overwrite with Berkeley DB