Re: VACUUM FULL doesn't reduce table size

From: Bill Moran <wmoran(at)potentialtech(dot)com>
To: pinker <pinker(at)onet(dot)eu>
Cc: pgsql-general(at)postgresql(dot)org
Subject: Re: VACUUM FULL doesn't reduce table size
Date: 2015-03-09 16:19:21
Message-ID: 20150309121921.4e6c6c6f5503e65adc127ddb@potentialtech.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

On Mon, 9 Mar 2015 09:05:07 -0700 (MST)
pinker <pinker(at)onet(dot)eu> wrote:

>
> > So there are no longer any dead rows being left behind, right?
> >
> > Why are we still discussing this? Do you have some other question?
>
> There are no dead rows, but postgres still cannot reuse the space because of
> 3043947 nonremovable row versions ..
>
> INFO: vacuuming "my_table"
> INFO: "my_table": found 0 removable, 3043947 nonremovable row versions in
> 37580 pages
> DETAIL: 0 dead row versions cannot be removed yet.
> CPU 2.67s/1.59u sec elapsed 7.71 sec.
> Query returned successfully with no result in 8319 ms.

Given your weird description of the "snapshot" I wouldn't be surprised if
that instance of PostgreSQL had subtle corruption.

pg_dump that database, re-init it and reload the dump. Then recreate the
situation and see if the rows are sill nonremovable. I bet you $5.34 that
everything works fine after that, which would indicate that the folks
who made the snapshot didn't do it correctly.

--
Bill Moran

In response to

Browse pgsql-general by date

  From Date Subject
Next Message Joshua D. Drake 2015-03-09 16:19:23 Re: VACUUM FULL doesn't reduce table size
Previous Message pinker 2015-03-09 16:05:07 Re: VACUUM FULL doesn't reduce table size