Обсуждение: VACUUM on multi-CPU systems?

Does anyone have any thoughts on whether my database vacuum/full/analyze
might go faster on our 2-CPU system if I generated the VACUUM commands
into a script, e.g.,

    psql -a <<EOF
\t
\o /tmp/db_maint.$$
select distinct on (relname) 'VACUUM FULL ANALYZE "' || relname || '";'
from pg_class
where relkind='r'
\g
\o
EOF

(generates a script that looks like

  VACUUM FULL ANALYZE table1;
  VACUUM FULL ANALYZE table2;

etc.)

and then started two psql sessions each processing half of the file?

--
Jeff Boes                                      vox 269.226.9550 ext 24
Database Engineer                                     fax 269.349.9076
Nexcerpt, Inc.                                 http://www.nexcerpt.com
           ...Nexcerpt... Extend your Expertise

On Mon, 26 Aug 2002 12:04:23 -0400, Jeff Boes wrote:

> Does anyone have any thoughts on whether my database vacuum/full/analyze
> might go faster on our 2-CPU system if I generated the VACUUM commands
> into a script, e.g.,
>
>     psql -a <<EOF
> \t
> \o /tmp/db_maint.$$
> select distinct on (relname) 'VACUUM FULL ANALYZE "' || relname || '";'
> from pg_class
> where relkind='r'
> \g
> \o
> EOF
>
> (generates a script that looks like
>
>   VACUUM FULL ANALYZE table1;
>   VACUUM FULL ANALYZE table2;
>
> etc.)
>
> and then started two psql sessions each processing half of the file?

I didn't see any response to this, so I'll try just one more time and
then shut up ...


--
Jeff Boes                                      vox 269.226.9550 ext 24
Database Engineer                                     fax 269.349.9076
Nexcerpt, Inc.                                 http://www.nexcerpt.com
           ...Nexcerpt... Extend your Expertise

On Fri, Sep 06, 2002 at 10:14:14AM -0400, Jeff Boes wrote:
>
> I didn't see any response to this, so I'll try just one more time and
> then shut up ...

Sorry, I think I was still on vacation when I responded to this, and
I failed to change the Subject: header.  But you can see my response
at


http://groups.google.com/groups?q=VACUUM+on+multi-CPU+systems+group:comp.databases.postgresql.*&hl=en&lr=&ie=UTF-8&selm=20020826132354.I22309%40mail.libertyrms.com&rnum=2

(Nasty long line.  Sorry.)

A

--
----
Andrew Sullivan                         204-4141 Yonge Street
Liberty RMS                           Toronto, Ontario Canada
<andrew@libertyrms.info>                              M2P 2A8
                                         +1 416 646 3304 x110

On Fri, 2002-09-06 at 16:14, Jeff Boes wrote:
> On Mon, 26 Aug 2002 12:04:23 -0400, Jeff Boes wrote:
>
> > Does anyone have any thoughts on whether my database vacuum/full/analyze
> > might go faster on our 2-CPU system if I generated the VACUUM commands
> > into a script, e.g.,
> >
> >     psql -a <<EOF
> > \t
> > \o /tmp/db_maint.$$
> > select distinct on (relname) 'VACUUM FULL ANALYZE "' || relname || '";'
> > from pg_class
> > where relkind='r'
> > \g
> > \o
> > EOF
> >
> > (generates a script that looks like
> >
> >   VACUUM FULL ANALYZE table1;
> >   VACUUM FULL ANALYZE table2;
> >
> > etc.)
> >
> > and then started two psql sessions each processing half of the file?
>
> I didn't see any response to this, so I'll try just one more time and
> then shut up ...

I think that vacuum will be more rapidly disk-bound then CPU-bound.

So if you've got different spindles for your tables, go ahead and
perform multiple vacuum full's at the same time (even on a uniprocessor
machine)

The proof of the pudding is of course what you all know (and like !)
Tycho

--
Tycho Fruru                            tycho@fruru.com
"Prediction is extremely difficult. Especially about the future."
  - Niels Bohr