I'm having a weird issue with BGP peering in address-family ipv6 unicast from a ASR1002-F running 3.1.1S (15.0(1)S) to a 7200 running 12.2(33)SRD3 and another ASR1002-F running the same software, that I have setup in a lab.
When a route-map (regardless of the contents, even route-map XXX permit 10 doesn't work) is applied on the outbound to the ipv6 neighbor of the ASR with 3.1.1S, it seems to stop sending labels for the IPv6 prefixes, the labels for show bgp ipv6 unicast labels show nolabel/nolabel for the routes in question and debugging ldp doesn't show the labels being sent. The result of this is that the route shows up on the 7200 in BGP for ipv6 unicast, but the next hop is inaccessible, and the route doesn't go into the ipv6 table obviously. As soon as I remove the outgoing route-map and soft reset the session on outbound, the next-hop is no longer inaccesible and the labels are there if I do a show bgp ipv6 unicast labels. If I re-apply the route-map and soft reset the session on outbound again, the BGP session sometimes (but not always) resets with the following error: *Oct 7 02:36:29.970: %BGP-3-NOTIFICATION: received from neighbor 192.168.1.3 3/10 (illegal network) 1 bytes 98 And when it comes back up, the labels have disappeared and the problem has resurfaced. Even if the session does not reset, and no error is shown, it the symptons are same, the labels are not sent. This seems to be an issue with the ASR not the 7200, the same symptom is seen on a ipv6 peering to another ASR1002-F when a outgoing route-map is applied to the bgp neighbor. Note that this only seems to affect IPv6 unicast, I'm also testing vpnv4 and vpnv6 (using 6VPE) all of which work without issues, with the same peer-policy inherited (and therefore the same route-maps). The first release that supports 6PE according to the release notes (unless I misunderstood them) is 3.1.0S so there is no going back to 2.X, I could try 3.1.0S instead of 3.1.1S but that seems somewhat counter productive. Anyone seen this before, or have any suggestion on how I could get around this? Kind regards, -- Sigurbjörn B. Lárusson _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
