OK. Here it goes: Your sites 68.15.53.176/25 and 68.15.53.174/25 are on the same subnet. However, due to the cable architecture they cannot see each other directly. The upstream router (which is visible to the world as 68.9.8.22) has a private IP 10.4.56.1, doing proxy arp for all the hosts on that subnet.
*) When receiving packets from 68.15.53.174 destined for 68.15.53.176 the router detects that the incoming and outgoing interface is the same which triggers the ICMP redirect that you were seeing. In this case you can/must ignore them. *) The fact that your UDP-based traceroute doesn't work can be due to the firewalling rules that you might have on 68.15.53.176. One question though, where does the GRE tunnel you were talking about come into play here? Ramin On Mon, Apr 29, 2002 at 10:36:03PM -0400, Mark Orenstein wrote: > 68.15.53.174 and 68.15.53.176 are the connections to the Internet for two > schools. The subnet mask is 255.255.255.128. Both connections are via cable > modems, most likely on the same cable segment. 10.4.56.1 must be the Cox > Communications router on the head end. When I traceroute from either side to > the other, it shows up as 1st in the traceroute output. An interesting thing > is that both traceroutes do not complete successfully to the other end. > However, a traceroute -I completes in two hops. > > [root@allsrv01 root]# traceroute 68.15.53.176 > traceroute to 68.15.53.176 (68.15.53.176), 30 hops max, 38 byte packets > 1 10.4.56.1 (10.4.56.1) 8.714 ms 10.247 ms 9.723 ms > 2 * * * > 3 * * * > 4 * * * > 5 * * * > 6 * * * > 7 * * * > 8 * * * > 9 * * * > 10 * * * > 11 * * * > 12 * * * > 13 * * * > 14 * * * > 15 * * * > 16 * * * > 17 * * * > 18 * * * > 19 * * > [root@allsrv01 root]# > > [root@squidhs root]# traceroute -I 68.15.53.174 > traceroute to 68.15.53.174 (68.15.53.174), 30 hops max, 38 byte packets > 1 10.4.56.1 (10.4.56.1) 8.854 ms 7.689 ms 8.126 ms > 2 wsip68-15-53-174.ri.ri.cox.net (68.15.53.174) 21.487 ms 23.157 ms 15.164 > ms > [root@squidhs root]#
