From: | Robert Haas <robertmhaas(at)gmail(dot)com> |
---|---|
To: | Simon Riggs <simon(at)2ndquadrant(dot)com> |
Cc: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Pavan Deolasee <pavan(dot)deolasee(at)gmail(dot)com>, pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Set visibility map bit after HOT prune |
Date: | 2012-12-20 15:55:27 |
Message-ID: | CA+TgmoaU6UtjaqVb4+kqrUaZ_pcDXVjMtoOEhwb_T93r+qse+A@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On Thu, Dec 20, 2012 at 4:58 AM, Simon Riggs <simon(at)2ndquadrant(dot)com> wrote:
> ISTM that if someone spots a block that could use cleanup, they mark
> the block as BM_PIN_COUNT_WAITER, but don't set pid. Then when they
> unpin the block they send a signal/queue work for a special cleaning
> process to come in and do the work now that nobody is waiting. Logic
> would allow VACUUMs to go past that by setting the pid. If we
> prioritised cleanup onto blocks that are already dirty we would
> minimise I/O.
I don't favor that particular signaling mechanism, but I agree that
there is quite a bit of potential utility in having foreground
processes notice that work (like a HOT prune, or setting the VM bit)
needs to be done and pass those requests off to a background process.
I'm hoping the new background worker framework in 9.3 will make that
sort of thing easier for people to play around with.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company
From | Date | Subject | |
---|---|---|---|
Next Message | Greg Stark | 2012-12-20 15:58:12 | Re: Parser Cruft in gram.y |
Previous Message | Robert Haas | 2012-12-20 15:53:32 | Re: Set visibility map bit after HOT prune |