Re: [COMMITTERS] pgsql: Properly set relpersistence for fake relcache entries.

Hi,

On Wednesday, September 26, 2012 07:57:06 AM Виктор Егоров wrote:
> I'm afraid I'm exactly in this situation now.
:(

> Last entry from the 9.1.6 recommended VACUUM (FREEZE, VERBOSE, ANALYZE)
It recommended doing a REINDEX first though? I guess you didn't do that?

> was: INFO: "meta_version_chunks": found 55363 removable, 32566245
> nonremovable row versions in 450292 out of 450292 pages
> DETAIL: 0 dead row versions cannot be removed yet.
> There were 588315 unused item pointers.
> 0 pages are entirely empty.
> CPU 2.44s/5.77u sec elapsed 2150.18 sec.
> INFO: vacuuming "pg_toast.pg_toast_16582"
>
> And here're are the locks held by the VACCUM backend:
> select
> oid,relname,relkind,relpages,reltuples::numeric(15,0),reltoastrelid,reltoas
> tidxid from pg_class
> where oid in (select relation from pg_locks where pid = 1380);
> oid | relname | relkind | relpages | reltuples |
> reltoastrelid | reltoastidxid
> -------+----------------------+---------+----------+-----------+-----------
> ----+--------------- 16585 | pg_toast_16582 | t | 16460004 |
> 58161600 |
> 0 | 16587
> 16587 | pg_toast_16582_index | i | 188469 | 58161600 |
> 0 | 0
> 16582 | meta_version_chunks | r | 450292 | 32566200 |
> 16585 | 0
>
> I will not touch anything and would like to get some recommendations on how
> to proceed.

On Wednesday, September 26, 2012 08:12:37 AM Виктор Егоров wrote:
> Forget to mention, that:
> - VACUUM is running on the master;
> - current state is unchanged for 20 hours.
I guess you cannot cancel the vacuum? Last time it was in a cycle without
checking interrupts inbetween.

Can you restart the server?

Greetings,

Andres
--
Andres Freund http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services