Re: more problems with count(*) on large table

On Fri, Sep 28, 2007 at 12:50:34PM -0400, Alvaro Herrera wrote:
>
> But he does have a WHERE condition.  THe problem is, probably, that the
> condition is not selective enough so the planner chooses to do a
> seqscan.

Or else the planner has a bad idea of how selective the condition is.
I've found that this sort of inconsistent selectivity on an indexed
column can be because the column statistics aren't good enough.  You
can try ALTER TABLE...SET STATISTICS to see if it helps.

A

--
Andrew Sullivan  | ajs@crankycanuck.ca
This work was visionary and imaginative, and goes to show that visionary
and imaginative work need not end up well.
        --Dennis Ritchie