"Francisco Reyes" <lists(at)stringsutils(dot)com> writes:
> Given that memory grows over time I am beggining to wonder if it is some
> type of memory leak.
Are there any AFTER triggers (including foreign key constraints) on the
table being inserted into? If so the list of pending trigger events
might be your problem.
If you can get Postgres to report an actual out-of-memory error (as
opposed to crashing from OOM kill) then it should dump a memory usage
map into the postmaster log. Looking at that would be informative.
regards, tom lane