"Josh Berkus" <josh(at)agliodbs(dot)com> writes:
> 1. INDEXES: I discovered, the hard way, a peculiar problem. If you drop
> and re-create a table within the same transaction (in a function, for
> example) the indexes do not get dropped completely.
Kinda hard to believe; even if the old indexes were still around, they
wouldn't be considered to apply to the new table. I think the problem
is something else. Can you provide a reproducible example of what
you're seeing?
> runs in about 2 seconds right after a VACUUM.
> Unfortunately, 6 hours after a VACUUM, the query bogs down.
What has been changing in the meantime?
regards, tom lane