dlw@VIOLET.BERKELEY.EDU (David Wasley) (04/08/88)
Has anyone else seen this? It is only a problem in this case because the host has only one address: its end of the pt-to-pt link. It advertises a "host route" (running 4.3bsd) only. David ---- From CLIFF@cmsa.berkeley.edu Thu Apr 7 09:54:16 1988 Date: Thu, 07 Apr 88 09:54 PDT From: CLIFF@cmsa.berkeley.edu (Cliff Frost) Subject: Proteon 7.4 To: KARELS@okeeffe.berkeley.edu, rwh@ucbarpa.berkeley.edu Cc: DLW@violet.berkeley.edu Hello, In the process of upgrading all our p4200 gateways to release 7.4 and I noticed that there is a bug fix that can create a problem for us in this release. It has to do with pt-to-pt routes. Proteon has never supported host routes. What they used to do when they got a RIP packet with a route to a host was turn that into a route to that host's entire network (or subnet). This seems obviously the wrong thing to do in general, but it worked ok for us because all the 140 subnet links are on one side of a p4200 and there was only one 186 subnet link. Now Proteon ignores host routes (better behaviour than before, I guess), and all of a sudden everyone on the far side of our p4200 couldn't get to Mills (128.32.186.2). So, I had to put a static route to the 186 subnet in our p4200. You may have similar problems with the proteon router onto the 137 subnet and various 140 subnet host routes. If so, let me know and I'll add static routes to that box as well. Cliff ps If you have a better idea on how to resolve this, I'll be more than happy to hear it.
braden@VENERA.ISI.EDU (04/08/88)
The solution to this may be bureacratic, not technical. When Hedrick's RIP RFC F*I*N*A*L*L*Y gets published (and I have reason to believe that will be soon), we will have a protocol document for RIP. Either it legitimizes host routes, or it doesn't (I don't know the answer). If it does, we can tell Proteon they are not implementing RIP. If not, we can tell whoever is sending that host route that they are broken. Bob Braden