The full causes an EXCLUSIVE lock, and therefore may cause the inserts to
block.
With a LAZY (I.E. Non-Full) Vacuum, we don't grab the exclusive lock, and
the free space
Is added to the Free-Space Map (FSM).
The PGDG (PostgreSQL Global Development Group) recommends lazy vacuums if at
all possible.
LER
--
Larry Rosenman http://www.lerctr.org/~ler
Phone: +1 512-248-2683 E-Mail: ler@???
US Mail: 430 Valona Loop, Round Rock, TX 78681-3893
-----Original Message-----
From: Sartorelli, Kevin [mailto:Kevin.Sartorelli@openpolytechnic.ac.nz]
Sent: Wednesday, September 20, 2006 4:13 PM
To: Larry Rosenman; Tom Kistner
Cc: exim-users@???
Subject: RE: [exim] Exilog database size
Thanks for the advice.
I have patched exilog_sql.pm as below (and added a FULL to reduce the DB
size) and re-ran the clean up. All is looking much better :-)
Cheers
Kevin
-----Original Message-----
From: Larry Rosenman [
mailto:ler@lerctr.org]
Sent: Thursday, 21 September 2006 1:43 a.m.
To: 'Tom Kistner'
Cc: Sartorelli, Kevin; exim-users@???
Subject: RE: [exim] Exilog database size
Tom,
yes, the patch I have in the FreeBSD port is to change (for
PostgreSQL) the OPTIMIZE TABLE to VACUUM ANALYZE.
--
Larry Rosenman http://www.lerctr.org/~ler
Phone: +1 512-248-2683 E-Mail: ler@???
US Mail: 430 Valona Loop, Round Rock, TX 78681-3893 -----Original
Message-----
From: Tom Kistner [mailto:tom@duncanthrax.net]
Sent: Wednesday, September 20, 2006 2:31 AM
To: Larry Rosenman
Cc: 'Sartorelli, Kevin'; exim-users@???
Subject: Re: [exim] Exilog database size
Larry Rosenman wrote:
> What are your FSM settings for PostgreSQL? Did you update the cleanup
> routine to actually Do the vacuum? (I'm the FreeBSD port maintainer,
> and the patch is in the FreeBSD ports collection).
Larry is probably right. At least I don't remember doing anything related to
a "Vacuum" in the postgres SQL stubs (maybe this is the equivalent of MySQLs
"optimize table"?).
/tom
