> > As it was getting unmanagable, I finnally hacked rpcbind code to get it
> > working again. Some code tries to avoid answering the same request
> > twice
> 
> usually a re-send from the client is either a timeout without
> answer or the packet got lost (from the point of view of the
> client). does your sentence imply that per default rpcbind has
> no tolerance for loss of packets ? (that's a tcp-handled matter,
> no ?)

It's even worse than that: once rpcbind corrupted its internal, it will
not serve at all the given client. It's just blacklisted, ignored, or
whatever you want.

Now I have no idea what cause the problem. A packet loss might be a good
path to explore. The switches in my compter lab are very weak, they
loose many packets.

-- 
Emmanuel Dreyfus
Il y a 10 sortes de personnes dans le monde: ceux qui comprennent 
le binaire et ceux qui ne le comprennent pas.
manu@netbsd.org