Edgar Fuß <Edgar.Fuss@bn2.maus.net> writes:

> But anyway, even if increasing NMBCLUSTERS really solves the problem
> (and not just makes the server take longer before it happens again)

It may be the case, but I haven't found out what's going on yet.
Either:
        1) we changed something and default number of mbufclusters
        isn't enough;
        2) we have an mbuf leakage.

Either way, both are strange and I have no idea what could have
changed..

> I would have expected a message suggesting to raise that option
> rather than ARP to lock up.

If it turns out that the best way to fix this problem is to increase,
NMBCLUSTERS, I'll change the message myself.

Thanks.
-- 
  Rui Paulo			<rpaulo@{NetBSD{,-PT}.org,fnop.net}>