Re: Some Autovacuum Questions

Поиск
Список
Период
Сортировка
От Alvaro Herrera
Тема Re: Some Autovacuum Questions
Дата
Msg-id 20080212153520.GL14683@alvh.no-ip.org
обсуждение исходный текст
Ответ на Some Autovacuum Questions  ("Thomas Chille" <thomas@chille.de>)
Список pgsql-general
Thomas Chille wrote:

> My 1. question is,
> if the known bugfixes for autovacuum after release 8.1.4 addressing my
> depicted issues?

Not directly, but keep reading.

> We are still using 8.1.4 because a database upgrade for us and our
> product is a hefty step wich involves a lot of customer databases. But
> if it could help we consider to upgrade to 8.1.11 or 8.3. What would u
> suggest?

You will be really happy when your database stops working because of a
bug in autovacuum which was fixed in 8.1.6.  I should add that perhaps
autovacuum is not processing your databases at all because it is stuck
on processing template0 due to that bug -- this could cause dead rows to
accumulate.

I'm not sure why you say it's a hefty step to upgrade to 8.1.11.
Perhaps you're not up to speed on our upgrade procedures.

On the other hand, perhaps this bug hasn't bit you yet.  But I can
promise, if you don't upgrade it will.

> My 2. questions is,
> if i possible configured something improper?

If you didn't tune autovacuum at all, yeah, perhaps it would be good to
adjust the settings somewhat.


> My 3. question is,
> if it possible to read the vaccuming or analyzing state of a given
> table manually? Are there any indicatores in statistic tables, wich
> the autovacuum demaon is using too wich can show me the progress of a
> running autovacuum?

Yes, they are kept in pg_stat_all_tables.  (Not all autovacuum numbers
are displayed however).

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

В списке pgsql-general по дате отправления:

Предыдущее
От: Chander Ganesan
Дата:
Сообщение: Re: Some Autovacuum Questions
Следующее
От: Alvaro Herrera
Дата:
Сообщение: Re: deadlock while re-indexing table