"Rafael Domiciano" <rafael.domiciano at gmail.com> wrote:
> 
> Bill,
> You said that VACUUM FULLs are not an effective method of ongoing
> maintenance. But VACUUM FULL is used by more than 5 years, so VACUUM is a
> culture in the company. The DBA function is new in the company, and I have
> been reading everything I find about Postgres, DB, Slony, Maintenance, and
> so on.
> I would like to know if you could pass me a forum or a list of messages on
> Postgres or maintenance in DB.
> If you can, I would be very grateful.

http://www.postgresql.org/docs/8.3/static/routine-vacuuming.html

> People,
> I'll give a new search in the documentation to find something about slony's
> vacuum and how to stop that.
> The problem is that I have an external application, and when the Slony's
> vacuum is running the application begins to slow down and give time-out.
> I think necessary to run the vacuum when it asks, but while the company
> decides not replace the hardware, I have to do that anyway. However, they
> are warned that this could affect the replication.

You're still going down the wrong road.  While it's possible that your
hardware is inadequate for the job, it's also possible that the same person
who thought VACUUM FULL was a good idea also thinks lots of other incorrect
things.  Before trying to redesign Slony and operating counter to the
advice of the very smart people who designed it, I would do a basic audit
of your PostgreSQL config settings and tune the server.  It's quite possible
that a little bit of postgresql.conf tuning will bring this thing back up
to par.  See:
http://www.postgresql.org/docs/8.3/static/admin.html

You may also want to join the pgsql-perform mailing list.

I'd be willing to bet that if you disable Slony's vacuum, you'll
make the situation worse.  Please try it, I'd love to hear your results.

-- 
Bill Moran
Collaborative Fusion Inc.

wmoran at collaborativefusion.com
Phone: 412-422-3463x4023