Running out of memory...

Hi All,

I am using PostgreSQL version 6.5.1 on x86 Solaris 2.6.

Now, I have already tweaked my shared mem settings as was
recommended
on this thread in the past.

I am also starting the postmaster with a -N 128 -B 256 -i.

The problem is this, I am getting numerous "unexpected EOF from
client" messages, and the postmaster keeps spawning (albeit slowly)
until I hit the max number of processes, and I get the expected "too
many clients already".

Does anyone know why the postmaster would need to spawn so many
backends? Has anyone seen this problem or a similar one on their Solaris
box?

Thanks in advance,

-Kevin