masks do not need to match on a virtual link for obvious reasons, those being that one cannot be certain of the end points. I suppose that in practical terms, one should always use /30's on serial links, and thus the end point masks would always match, but who can ever tell? I suppose it is possible that one end of a virtual link could be an ethernet or a token ring interface, and the distant end a serial interface, and thus it would be likely that masks do not match. ( and yes I know that in the case of Cisco, anyway, that the RID is the end point, and RID's don't have masks anyway. ) BTW, a virtual link hello has the v-bit set - it is that which determines that the packet is for purposes of a virtual link.
the point to point link masks not having to match is interesting. one of these days I'll have to set something up in the lab, just to see. not generally being one to deliberately setting things up incorrectly, I sometimes miss out on these kinds of curiousities. Chuck ""Priscilla Oppenheimer"" wrote in message [EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > With regards to the items below, I would say that Cisco follows the RFC, > but just describes the issues a little differently. See comments below. > > At 04:28 PM 4/16/02, Kane, Christopher A. wrote: > > >It's within the Hello protocol that there are certain criteria that must be > >met. ACCORDING TO CISCO they are: Hello/Dead Interval, Area ID, Stub Flag > >and Authentication [method and password]. So, I wanted to see what RFC 2328 > >had to say about it. I also checked John T. Moy's book, Anatomy of an > >Internet Routing Protocol. In both of those sources I find that the > >following must match: Network mask, HelloInterval and RouterDeadInterval and > >the E-bit of the Options Field. The exception being the Network mask > >(depending on the Network Type in use). > > > >RFC states: > >HelloInterval > > Cisco says this must agree also. > > >RouterDeadInterval > > Cisco says this must agree also. > > >Network Mask > > The RFC says to ignore this on point-to-point networks and on virtual > links. Maybe Cisco just doesn't mention it because it's not a rule that > always applies. > > >E-bit of Options Field (Area capable of processing AS-external-LSAs) > > That's what Cisco calls the stub flag I bet. > > > >Cisco implementation: > >Hello/Dead Interval > >Area ID > > The RFC covers this too, but in the general discussion, not just in the > discussion of Hellos. The Area ID in an OSPF packet must match the area of > the receiving interface (except in the case of virtual links, in which case > it must indicate the backbone). > > >Stub Flag > >Authentication Method/password > > The RFC says this must agree on every OSPF packet. It just doesn't > specifically mention that it must agree on Hello packets. > > > >I realize vendors have the choice of how closely they follow an RFC. > > If the RFC says "must" then a vendor must do what it says. It's only when > it says "should" or in grey areas where the authors didn't make something > clear that you run into problems. > > > I'm > >just trying to make sure I understand the protocol for what it is and for > >how Cisco deploys it. Can someone experienced with this protocol check my > >understanding? > > > >-chris > ________________________ > > Priscilla Oppenheimer > http://www.priscilla.com Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=41663&t=41647 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
