> Since our nightly window of relative inactivity continues to shrink, I'd > like to vacuum/analyze only the tables that autovacuum thinks need it, > all at once, and only when I tell it to (via cron). I thought as you did and posted a thread, where I proposed a C function smartvacuum() which you can call to vacuum with the same criteria as autovacuum on demand. http://archives.postgresql.org/pgsql-patches/2006-10/msg00131.php But that was hardly accepted http://archives.postgresql.org/pgsql-hackers/2006-10/msg01190.php and the discussion are gone over there: http://archives.postgresql.org/pgsql-hackers/2006-10/msg01225.php There're some cases where the whole database is not needed to be vacuumed, right? Anyway, though smartvacuum() was not implemented as I could find another way, you maight be able to implement it by yourself. Or, any way to catch row level statistic from SQL or pgsql function should be prepared on future version?? Regards, Hitoshi Harada > -----Original Message----- > From: pgsql-general-owner@xxxxxxxxxxxxxx > [mailto:pgsql-general-owner@xxxxxxxxxxxxxx] On Behalf Of Glen Parker > Sent: Friday, November 10, 2006 11:05 AM > To: Postgres General > Subject: Re: [GENERAL] AutoVacuum on demand? > > Richard Broersma Jr wrote: > >> Cron yes, vacuumdb no. I would like to AUTO vacuum periodically. > > > > Just curious, what is auto_vacuum going to give that vacuumdb wont? > > > > Hmm, I thought this would be obvious. Vacuumdb just issues "VACUUM" > commands. > > I would like the get the functionality if autovacuum so that I don't > waste cycles on tables that don't really need vacuuming. However, we've > had bad luck letting autovacuum run on its own terms, so we've reverted > to the old nightly vacuums. > > Since our nightly window of relative inactivity continues to shrink, I'd > like to vacuum/analyze only the tables that autovacuum thinks need it, > all at once, and only when I tell it to (via cron). > > Make more sense? > > -Glen > > ---------------------------(end of broadcast)--------------------------- > TIP 9: In versions below 8.0, the planner will ignore your desire to > choose an index scan if your joining column's datatypes do not > match
