Shaun Thomas wrote:
> Ok, that might explain it, then. We did in fact just upgrade from 8.2 to
> 9.1 about 2 weeks ago. And no, I didn't do a VACUUM FREEZE, just a
> VACUUM ANALYZE to make sure stats were ready. I'm still a little
> uncertain what the tangible difference is between a FREEZE and a regular
> VACUUM. I get that it sets freeze_min_age to 0, but why does that even
> matter? Is 50M out of 2B not good enough? Every VACUUM knocks the
> counter back to the minimum, so I guess I don't get the justification
> for magically forcing the minimum to be lower.
You might find this section of the docs illuminating:
http://www.postgresql.org/docs/9.2/interactive/routine-vacuuming.html#VACUUM-FOR-WRAPAROUND
-Kevin
