Tom Martin wrote: > Robert, > > I believe that your diagram should reflect R1's serial interface to R2 > as s0/1 instead of s0/0. This caused me some confusion in trying to > figure out the configs. Actually, there is still some confusion given
Sorry for that - this was copy-pasting error (there's a whole bunch of other stuff running on the routers, I had to filter these out). The addressess are unique, obviously - if they were'nt then the router would drop (as unroutable) the packet with src_IP equal to IP of one of its interfaces - so the RIP update would have got no chance to enter the RIP process. > You make a strong argument that a more logical interpretation would be > to use the local IP address 172.16.66.1/25 to interpret the route since > it is the only IP address that is on the same subnet as the sending > router (since the other IPs configured on the link should, based on > normal IP rules, require another router to communicate with the sender). > All documentation I've come across and configuration I have done > indicates that the receiving router validates the update based on major > network only, and then uses the mask of the locally configured address > of that network to interpret the incoming networks. So, technically, > interpreting the route as 172.16.77.0/29 isn't "wrong" -- it's just one > of 3 possible ways of interpreting the advertised network. That's correct - all choices, according to many written sources, are perfectly correct. But the router has to break a tie - in this case longest subnet mask was chosen. I'm still curious if this behaviour is defined somewhere or this is Cisco-feature IOS-dependent one. For reference: below is an algorithm, hopefully complete, for classfull processing of RIP updates compiled by me from various sources and documents including Doyle and Zinin. I had to add 'longest' to 'apply_mask_of_incoming_interface' based on results of testing this issue. Receiving (update): if (major net of update the same as of incoming interface ?) { NO: if (any subnets of major net of update already exist in route table known from other interfaces ?) { YES: discard( update ); exit(); NO: apply_classfull_mask( update ); populate_rt( update ); exit(); } YES: if (there are any bits in host portion) { NO: apply_longest_mask_of_incoming_interface( update ); populate_rt( update ); exit(); YES: apply_32_mask( update ); populate_rt( update ); exit(); } } Sending (update): if (subnet of update in the same major net as outgoing interface) { YES: if (subnet mask is the same as subnet mask of sending interface ?) { NO: if (the update is a host route ?) { YES: send( update ); exit(); NO: discard( update ); exit(); } YES: send( update ); exit(); } NO: summarize_classfull( update ); send( update ); exit(); } > > I'm curious as to whether your configuration works at all given the > next-hop address (172.16.66.1) is also a valid IP address on R2. Are > you able to ping 172.16.200.1 from R2? It seems to me that R2 should be It will drop packet as unroutable. robert, -- Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=57192&t=57049 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]