Subject: Re: NetBSD netbooting (was "Re: UV3100 SCSI")
To: Chris Craft <>
From: Brian D Chase <>
List: port-vax
Date: 06/16/1998 09:27:03
On Tue, 16 Jun 1998, Chris Craft wrote:
> At 23:11 6/15/98 -0400, entropy wrote:
> >>From: (Michael Sokolov)

> >>[...] It seems to
> >>me that the people who have designed NetBSD's netbooting mechanism have
> >>assumed that everyone is working on a "fake" network with 10.x.x.x-style
> >>IPs, no connection to the Internet, and no one besides themselves working
> >>on it. [...]
> >
> >In what way did you find NetBSD's netbooting mechanism deficient?
> >
> >I've netbooted NetBSD/vax, NetBSD/pmax, and NetBSD/i386.  All systems
> >were netbooted while connected to "real" networks (both the Internet
> >and a corporate LAN, on different occasions).  I never found any need
> >at all to use a "fake" network to boot any of my systems.
> I also had little trouble netbooting 1.3 on a real campus network with a
> live internet connection and lots of little lab rats hammering yahoo
> chat.  The only difficulties I found were with Linux' apparent
> FUBAR-edness with net3. 

The netbooting facilities were definitely not designed with the reserved
10 class A in mind, they are completely generic.  The 10 network just
happens to be a nice one (I mean if you're going to go for a reserved
network, you might as well grab the biggest one).  It's also a good idea
to use the reserved net addresses in documentation in case a novice takes
the use of the IPs literally.  The breadth of damage one can do is limited
by using the reserved addresses.

As far as problems netbooting go... it is likely that people may
experience difficulties in environments which make use of intelligent
ethernet switches or if you're trying to netboot from a machine on a
different network (through a router).  MOP being a broadcast protocol,
you'll have to ensure that it's packets get broadcast to the appropriate
location.  It's also quite possible that in some scenerios certain types
of packets may get filtered out.  You could have an ethernet switch which
only passes on allowed protocols and MOP is sort of an oddball one in this
day and age. 

Still, it's fairly easy and inexpensive to get a small dumb hub to setup
multiple machines within a single ethernet segment.

Also... pointing back toward the first topic in this thread.  I'll be
trying out Sokolov's kernel as soon as get a chance.  It would be nice to
be able to at least boot from a SCSI device.  This week's going to be a
busy one for me so I probably won't get a chance until this weekend

Brian "JARAI" Chase | | VAXZilla LIVES!!!