Jeff Frost <jeff@frostconsultingllc.com> writes:
> On Thu, 2 Nov 2006, Tom Lane wrote:
>> This seems pretty darn weird. I am wondering about corrupt indexes ---
>> can you find the indicated key in either table if you set
>> enable_indexscan and enable_bitmapscan to 0?
> test_tracking=# begin;
> BEGIN
> test_tracking=# set enable_seqscan TO false;
> SET
> test_tracking=# set enable_bitmapscan to false;
> SET
That's not actually the test case I asked for ...
> Nope, I guess I didn't \d visit..thought I did...it's below. I'm going to
> guess that there is index corruption and a reindex page_view will fix it. Do
> you want me to gather any information in case this is a reproducible bug
> before I issue the reindex?
I'm thinking the same --- please save a physical copy of
page_view_visit_idx before you reindex it, and if that fixes the
problem, please send me copies of both the before and after states
of the index.
regards, tom lane