From: | Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> |
---|---|
To: | Hannu Krosing <hannu(at)skype(dot)net> |
Cc: | Qingqing Zhou <zhouqq(at)cs(dot)toronto(dot)edu>, pgsql-hackers(at)postgresql(dot)org |
Subject: | Re: Autovacuum in the backend |
Date: | 2005-06-15 12:49:22 |
Message-ID: | 200506151249.j5FCnMq14570@candle.pha.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general pgsql-hackers |
Hannu Krosing wrote:
> On K, 2005-06-15 at 13:41 +0800, Qingqing Zhou wrote:
> > "Neil Conway" <neilc(at)samurai(dot)com> writes
> > >
> > > Wasn't the plan to rewrite pg_autovacuum to use the FSM rather than the
> > > stats collector?
> > >
> >
> > I don't understand. Currently the basic logic of pg_autovacuum is to use the
> > pg_stat_all_tables numbers like n_tup_upd, n_tup_del to determine if a
> > relation need to be vacuumed. How to use FSM to get these information?
>
> One can't probably use FSM as it is, as FSM is filled in by vacuum and
> this creates a circular dependency.
>
> But it would be very nice to have something _similar_ to FSM, say DSM
> (dead space map), which is filled in when a tuple is marked as "dead for
> all running backends", which could be used to implement a vacuum which
> vacuums only those pages, which do actually contain removable tuples.
Yes, those are step five. The TODO list has:
* Auto-vacuum
o Move into the backend code
o Scan the buffer cache to find free space or use background writer
o Use free-space map information to guide refilling
o Do VACUUM FULL if table is nearly empty?
--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073
From | Date | Subject | |
---|---|---|---|
Next Message | Devrim GUNDUZ | 2005-06-15 13:00:51 | Re: Consultants |
Previous Message | Csaba Nagy | 2005-06-15 12:07:44 | Re: Deadlock Problem |
From | Date | Subject | |
---|---|---|---|
Next Message | Bruce Momjian | 2005-06-15 12:51:07 | Re: LGPL |
Previous Message | Zac | 2005-06-15 12:29:27 | process crash when a plpython function returns unicode |