At 01:00 PM 3/26/99 +0100, Matthias Drochner wrote:
>While I'm still clueless about VAXen (see below), I've changed
>the diskless startup some weeks ago so that it should clean
>up interface addresses and routes if the mount fails.
>Perhaps you can try a newer kernel and report whether it
>works for you.

I would love too, getting to a newer kernel has been somewhat more
difficult than I anticipated. I'm currently running the 1.3K kernel that
Michael built and completely off NFS disks, this avoids the MSCP driver bug
for now. The downside of using this kernel is that ps(1) no longer works.
(as well as anything that looks at kvm I believe). This is no doubt due to
the fact that I've only got the 19981201 snapshot binaries available until
I can make the world again.

As soon as I can understand the subtleties of the NetBSD make process I
believe the system is stable enough under NFS to get a complete crankturn
of the build. So far however 'make build' only does the includes and
cleandirs, it doesn't build the system (yes MACHINE_ARCH and MACHINE are
both set to 'vax'). 

>DHCP is much better for diskless startup because it can pass
>all needed parameters, so it is the first tried for a good reason.
>It seems to me that your coexistance problems are rather caused
>bu a stupid, or misconfigured, or both, DHCP server. (Let me guess
>what OS it runs on... :-)

Actually, it runs on FreeBSD. 

The trick here is that it runs as part of a product I helped build called a
'OneGate' gateway. Its an easy to use internet server product that lets non
network geniuses set up a network. Up to now we've not needed to provide
other options in the DHCP response messages :-) Needless to say, this will
be fixed but in the mean time, my suggestion of trying revarp first still
makes sense.

Cheers,
--Chuck McManis