On Dec 6, 2005, at 5:03 PM, Ameet Kini wrote:
> table with only 1 index, the time to do a vacuum (without full)
> went down
> from 45 minutes to under 3 minutes. Maybe thats not bloat but thats
> surely surprising. And this was after running vacuum periodically.
I'll bet either your FSM settings are too low and/or you don't vacuum
often enough for your data churn rate.
Without more data, it is hard to solve the right problem.