Rafael Martinez <r.m.guerrero@usit.uio.no> writes:
> On Tue, 2006-05-16 at 11:14 -0400, Tom Lane wrote:
>> So much for that theory. If you copy the table (create table foo as
>> select * from tickets) and build a similar index on the copy, does the
>> behavior persist in the copy?
> The new table behaves well:
Did you check that you were getting the same indexscan plans there?
(If not, try ANALYZEing the copied table.) You probably were, but
just in case.
This is definitely pretty baffling. I'm getting to the point where
I want to step through the code with a debugger. I assume that's not
very practical on your live server. Would it be feasible at all to
get a physical copy of the database for testing? Alternatively, do
you know C and gdb well enough to try to debug it for yourself?
regards, tom lane