Re: pg_stat_get_last_vacuum_time(): why non-FULL?

On Monday, April 08, 2013 4:40 AM CR Lender wrote:
> On 2013-03-31 18:31, CR Lender wrote:
> > On 2013-03-28 20:44, Kevin Grittner wrote:
> >> CR Lender <crlender(at)gmail(dot)com> wrote:
> >>> I've read the manual more carefully now, and I can't see any
> mention of
> >>> what VACUUM does that VACUUM FULL does not. The point about extreme
> >>> maintainance is taken, but from what I read, VACUUM FULL should
> include
> >>> everything a normal VACUUM does.
> >>
> >> Prior to release 9.0 that is probably true.
> >
> > Hm, I can't find it, even in the manual for 9.2.
> > http://www.postgresql.org/docs/current/static/sql-vacuum.html
> >
> > If VACUUM FULL is just a more aggressive VACCUM (including writing
> new
> > data files), then I don't understand the "non-FULL" restriction in
> > pg_stat_get_last_vacuum_time()... unless that information is somehow
> > lost when table files are rewritten.
>
> I don't mean to be pushy, but I have a meeting with the admin of that
> database tomorrow, and it would be nice if I had something concrete to
> tell him. I still don't know what it is that VACCUM does but VACUUM
> full
> doesn't do. There's nothing in the manual about that.

One of the important difference is that during the time VACUUM FULL is operating on a relation,
no other operations will be allowed on that relation. Most of admin care about this point, because
they don't want to stop operations for background garbage collect.
VACUUM FULL is only done in rare cases when the relation size has grown too bigger than it's actual
Contents.

With Regards,
Amit Kapila.