We have a 3000/600 running 3.2C configured for 736 maxusers - 288 meg of
memory with 582 meg of swap configured.
It normally runs something between 50 and 100 simultaneous users
with something like 280 (upwards) processes.
Twice now in recent weeks we have seen a situation where anyone logged in
or (apparently) any other daemon continues working, but new telnet users
are confronted with a message to the effect "all ports full."
Normally we run with about 50% free swap, but last evening when this
problem occurred we seemed to have a big spike in swap utliziation.
Theoretically, there are a plenty of ptys, but that's what it acts like -
out of ptys.
The system is a heavily used student mail system so when a problem like
this occurs, it tends to be rebooted "immediately" rather than take time to
hunt for the subtle problem.
What I'm looking for is
1) What does "all ports full" - really mean?
2) anybody seen this symptom before and have an idea what
parameter/resource is out of wack for a large multi-user system.
T.T.F.N.
William H. Magill
Information Services and Computing (ISC) University of Pennsylvania
Internet: magill_at_isc.upenn.edu magill_at_acm.org
magill_at_upenn.edu
http://pobox.upenn.edu/~magill/
Received on Wed Dec 06 1995 - 16:30:24 NZDT