I _think_ I might have figured out what's going wrong, but I can't
out how to solve for it. (Please feel free to tell me that what I
think is
wrong is wrong, as well, if that is the case...)

To redescribe: I have two private nics on a BM server - 10.1.x.x is my
network w/ novell servers and a w2K DHCP server. I'm setting up a new

10.2.x.x segment on the other card to serve students bringing in their
laptops, but don't want to dedicate a server to that segment. I've
bootpfwd on the BM server to route DHCP requests from the 10.2.x.x to
DHCP server on 10.1.x.x (The DHCP server is correctly configured with

In testing, requests get through to the DHCP server, but half of the
the offered leases are from the 10.1.x.x scope.

What dawned on me is that my testing machines are machines that used
to live
on the 10.1.x.x network, and what I am thinking is that when a DHCP
requests a DHCP lease, it first asks for the last IP address it had.
even though I have a machine on the 10.2.x.x segment and it's request
routing through the bootfwd server, it is asking for a 10.1.x.x
My guess, then, is that if that address is still available in the
scope the DHCP server happily offers it back. But since it is now
trying to
send a 10.1.x.x address to a machine on the 10.2.x.x segment of the BM

server, everything grinds to a halt.

Sound likely?

If so, my question is what I can do? True, when students bring in
their own
machines they won't have any previous 10.1.x.x addresses to ask for,
and so
everything should work. But I'd still like to be able to test without
stumbling block. Is there any way to clear the client's memory of
what its
previous address had been? Any way to request a brand new, fresh DHCP


Thanks for any thought!