From: | "Christopher Kings-Lynne" <chriskl(at)familyhealth(dot)com(dot)au> |
---|---|
To: | "Tom Lane" <tgl(at)sss(dot)pgh(dot)pa(dot)us>, "Bruce Momjian" <pgman(at)candle(dot)pha(dot)pa(dot)us> |
Cc: | "Nicolas Bazin" <nbazin(at)ingenico(dot)com(dot)au>, <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Postgresql backend to perform vacuum automatically |
Date: | 2002-03-05 06:03:52 |
Message-ID: | GNELIHDDFBOCMGBFGEFOKEKACBAA.chriskl@familyhealth.com.au |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
> > Yes, makes sense to me, especially now that we have a nolocking vacuum.
> > Tom, do you have any ideas on this?
>
> There's a TODO item about this already.
>
> * Provide automatic scheduling of background vacuum (Tom)
I think an good system would have some parameters something like this in
postgresql.conf:
vacuum_update_threshold
= num of operations that cause frags on a table before a vacuum is run, 0 =
no requirement, if followed by a percent (%) sign, indicates percentage of
rows changed, rather than an absolute number
vacuum_idle_threshold
= system load below which the system must be before a vacuum can be
performed. 0 = no requirement
vacuum_time_threshold
= seconds since last vacuum before which another vacuum cannot occur. 0 = no
requirement.
If all 3 are 0, then no auto-vacuuming is performed at all. There's
probably trouble if only the idle threshold is set to zero.
And the same for the 'analyze' command?
If they want it on a per-table basis, then they can just do it themselves
with a cronjob!
Chris
From | Date | Subject | |
---|---|---|---|
Next Message | Bruce Momjian | 2002-03-05 06:05:06 | Re: Postgresql backend to perform vacuum automatically |
Previous Message | Tom Lane | 2002-03-05 05:33:34 | Re: Postgresql backend to perform vacuum automatically |