Re: PATCH to allow concurrent VACUUMs to not lock each

Alvaro Herrera <alvherre@commandprompt.com>

From: Alvaro Herrera <alvherre@commandprompt.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Hannu Krosing <hannu@skype.net>, Bruce Momjian <bruce@momjian.us>, Neil Conway <neilc@samurai.com>, pgsql-patches@postgresql.org
Date: 2006-07-30T19:21:39Z
Lists: pgsql-hackers
Tom Lane wrote:
> Hannu Krosing <hannu@skype.net> writes:
> > Ühel kenal päeval, P, 2006-07-30 kell 14:11, kirjutas Alvaro Herrera:
> >> What was idea behind moving vac_update_relstats to a separate
> >> transaction?  I'm wondering if it's still needed, if it further enhances
> >> the system somehow, or your patch did something differently than what
> >> was applied.
> 
> > The part of transactions which actually modified the data (iirc it updates
> > relpages and reltuples in pg_class) is not safe to ignore by concurrent 
> > vacuum, say a vacuum on pg_class .
> 
> But that's done as a nontransactional update, or at least was the last
> time I looked, so there's no need to do it in a separate xact.
> 
> Knew I should have taken time to review that patch before it went in ...

Which one?  The one I applied doesn't have this change.  (You are still
more than welcome to review it of course.)

-- 
Alvaro Herrera                                http://www.CommandPrompt.com/
PostgreSQL Replication, Consulting, Custom Development, 24x7 support