Would anybody here know why a PE (Cat6500/Sup7203B running SXI1) would not decrement TTL for traces towards known (=has ARP entry) hosts on an interface?Would anybody here know why a PE (Cat6500/Sup720 running SXI1) would not decrement TTL for traces towards known (=has ARP entry) hosts on an interface?
I have no idea if the device has always behaved like that, but other PEs in the network behave as I expect. When I trace in the global table it works: # traceroute -I -w 2 -q 1 rhodd-fanout traceroute to rhodd-fanout (10.85.240.68), 30 hops max, 40 byte packets 1 vl3.aar-core-1.net.rm.dk (10.83.247.2) 0.273 ms 2 vl2275.aar-core-2.net.rm.dk (10.85.250.18) 0.290 ms 3 tge5-5.asthg-1.net.rm.dk (10.85.250.150) 0.351 ms 4 ge4-5.rhodd-1.net.rm.dk (10.85.250.154) 0.855 ms <---- 5 rhodd-fanout.net.rm.dk (10.85.240.68) 1.357 ms # The device in question is hop 4 here. Towards something in a VRF that doesn't exist, i.e. doesn't have an ARP entry: # traceroute -I -w 2 -q 1 10.235.248.20 traceroute to 10.235.248.20 (10.235.248.20), 30 hops max, 40 byte packets 1 vl3.aar-core-1.net.rm.dk (10.83.247.2) 0.266 ms 2 vl2276.aar-fws.net.rm.dk (10.85.250.22) 0.257 ms 3 vl2008.aar-fws.net.rm.dk (10.85.252.68) 0.621 ms 4 vl2277.aar-core-2.net.rm.dk (10.85.250.25) 1.314 ms 5 tge5-5.asthg-1.net.rm.dk (10.85.250.150) 1.492 ms 6 10.235.248.2 (10.235.248.2) 1.961 ms <---- 7 * 8 * ... Hop 6 is the device in question here. It sources the Time Exceeded from the VRF interface itself as normally. But when tracing something in a VRF and that does exist and has an ARP entry: # traceroute -I -w 2 -q 1 rhodd-dis-b traceroute to rhodd-dis-b (10.235.248.10), 30 hops max, 40 byte packets 1 vl3.aar-core-1.net.rm.dk (10.83.247.2) 0.244 ms 2 vl2276.aar-fws.net.rm.dk (10.85.250.22) 0.298 ms 3 vl2008.aar-fws.net.rm.dk (10.85.252.68) 0.829 ms 4 vl2277.aar-core-2.net.rm.dk (10.85.250.25) 1.534 ms 5 tge5-5.asthg-1.net.rm.dk (10.85.250.150) 1.411 ms 6 rhodd-dis-b.net.rm.dk (10.235.248.10) 4.931 ms # Where did it do this time? With "debug ip icmp" on the device I never see the expected "ICMP: time exceeded (time to live) sent to 10.83.247.20 (dest was 10.235.248.20)". I've checked the configuration and can find nothing to make this device different from any of the others. I'd be happy if anybody could hit me with a clue bat or tell me where to start looking... -- Peter _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
