"Jim C. Nasby" <jnasby@pervasive.com> wrote:
> And +1 on Rod's suggestion to make it more aggressive. I always drop the
> scale factor to at least 0.2 and 0.1 (though 0.1 and 0.05 don't seem
> unreasonable), and typically drop the thresholds to 200 and 100 (though
> again, lower is probably warrented).
The default fillfactors for index btree is 90%. So if we want to avoid
spliting of the leaf pages, vacuum scale factors should be less than 0.1
in cases where tuples are only updated randomly. I think threshoulds should
be less than PCTFREEs(=1-fillfactors) except ever-increasing tables.
This is a too simplified policy, but we probably need documentation for
the linkages between autovacuum and fillfactors.
Regards,
---
ITAGAKI Takahiro
NTT Open Source Software Center