> On Fri, 30 Nov 2007 19:34:02 -0300 > Alvaro Herrera <alvherre@alvh.no-ip.org> wrote: > > > Joshua D. Drake escribió: > >
> > > The multi-worker autovacuum is a great new addition to help part of > > > that problem (starvation) but it is not help against the other > > > (resource consumption, specifically IO). > > > > Huh, autovac will consume exactly the same amount of I/O as a > > user-induced vacuum, so I don't see your point. > > It can be determined "when" the I/O is used.
Future plans call for being able to change autovac parameters depending on time of day, so you will be able to decide that using autovacuum too.
> > Remember, we're not adding VACUUM SCHEMA in 8.2 so it doesn't make any > > sense to try to compare it against the old autovac. And you can bet > > that in 8.4 autovac will have even more goodies. > > I assume you mean 8.3, but that is certainly a valid point.
No, I meant 8.4 --- the feature set of 8.3 is already set on stone. So VACUUM SCHEMA has to compete with whatever we're able to do for the release following that one.
i think even when autovacuum is uber-perfect , the manual overriding commands like VACUUM and utilities like VACUUMDB will still exist for a long time to come, so there is a case of improving them if required. Maybe it would be just easier to add a schema switch to VACUUMDB, which can just query the catalogs to find which tables to vacuum, this way no parser changes are involved or any tinkering with the backend code.