Re: enabling autovacuum

[Jeremy Harris <jgh(at)wizmail(dot)org>]

Chander Ganesan wrote:
> Jeremy Harris wrote:
> > Version:
> >  PostgreSQL 8.2.4 on i686-redhat-linux-gnu, compiled by GCC gcc (GCC) 
> > 4.1.2 20070418 (Red Hat 4.1.2-10)
> >
> > We have one problematic table, which has a steady stream of entries
> > and a weekly mass-delete of ancient history.  The "bloat" query from
> > Greg Sabino Mullane (thanks to Greg Smith for pointing it out) returns:
> Inserts don't generate dead tuples, and AVD looks at obsolete tuples..  
> As such,  I wouldn't expect AVD to kick off until after you did a mass 
> delete...assuming that delete was sizable enough to trigger a vacuum.

Ah, that would explain it - thankyou.   So I need to retreat to
the question of why the weekly vacuum permits the observed bloat.
Any ideas?   More information that I could gather?

Thanks,
  Jeremy