> It's not the sort. There's a memory leak in the system somewhere. Can
> you post a reproducible test case? (If you can do it in the next couple
> of days, there's a chance the fix will make it to 8.3.4).
I'm not sure how to go about creating a test case for this. I have had a
nagging suspicion about a memory leak somewhere in this system (starting
back with this thread:
http://archives.postgresql.org/pgsql-general/2008-07/msg00048.php) but I
really don't have any ideas on were to even begin with creating a test
case. I did have a small test case with that thread
(http://archives.postgresql.org/pgsql-general/2008-07/msg00063.php) but
that was based on my guess about the XML stuff causing the leak. I have no
real proof of this, however, nor do I have any idea of a way to narrow
down the problem.
-- m@