On Thu, 4 Dec 2008, Matthew Wakeling wrote:
>> http://archives.postgresql.org/pgsql-bugs/2008-11/msg00180.php
>
> Thanks, that does explain everything.
Oh right, yes. It explains everything *except* the fact that the backend
is still holding onto all the RAM after the query is finished. Could the
fact that we SIGINTed it in the middle of the query explain that at all?
Matthew
--
I'd try being be a pessimist, but it probably wouldn't work anyway.