Re: Out of memory on SELECT in 8.3.5

"Matt Magoffin" <postgresql.org@msqr.us> writes:
> Here's the output of ulimit -a by the "postgres" user the database is
> running under:
> ...
> I think this means it does not have an artificial memory limit imposed,

Agreed, that ulimit isn't reflecting any such limit, but is that really
the same environment the postmaster gets started in?  I wouldn't trust
a system startup script to be launched in the same environment that a
login shell gets.  You might try adding
    ulimit -a >/tmp/something
to the startup script to find out for sure.

            regards, tom lane