Tom Lane wrote:
> Guy Thornley <guy@esphion.com> writes:
> > Below is a patch for the lazy vacuum. It implements a simple I/O throttle so
> > boxen arnt killed for hours a day when VACUUM runs.
>
> Wasn't this idea tried and rejected already? You haven't given us any
> information about actual performance.
What would be more valuable would be if VACUUM ran only when it know it
could reap a certain number of tuples, or there where FSM need for those
tuples.
--
Bruce Momjian | http://candle.pha.pa.us
pgman@candle.pha.pa.us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073