[c-nsp] Troubleshooting interface errors that travel with a REP topology change

[Jason Lixfeld]

I hope this ascii stuff works :|

I have two 7600s and four ME3400s in a couple of REP segments.


  +------------------------------------------------te8/2-----------------------------------------------+
  |                                                                                                    |
  +(ep)                                                                                                +
76001+(ep)-Gi9/13-Gi0/1-+ME1+-Gi0/2-Gi0/1-+ME2+-Gi0/2-Gi0/1-+ME3+-Gi0/2-Gi0/1-+ME4+-Gi0/2-Gi9/13-(e)+76002
  +(e)                                                                                                 +
  |                                                                                                    |
  +-----------------------------------------------gi9/20-----------------------------------------------+

- The 7600 Gi9/13 chain is segment 13.
- The 76001 Te8/2 76002 Gi9/20 72001 ring is segment 1024.
- (e)=segment edge port
- (ep)=segment edge primary port
- Segment 1024 is the actual ring.
- Segment 13 sends REP stcns to segment 1024.
- REP alternate port for segment 13 is ME2 Gi0/2.

What I'm seeing is a bunch of input interface errors on Gi9/13 and Te8/2 on 76002.
When I fail the REP segment at ME3 Gi0/2 forcing ME3 traffic via 76001, the errors move	to Gi9/13 on 76001.  The errors that were reported on Te8/2 don't follow to 76001.
When I return the segment back to normal, blocking at ME2 Gi0/2, errors return to Gi9/13 & Te8/2 on 76002.

It seems to me that the issue clearly has something to do with ME3, but I can't for the life of me figure out how to go about troubleshooting this.  I can't imagine that it could be layer 1 related.  Maybe ME3 is borked?  The errors don't seem to coincide with any excessive traffic along that link.  Traffic across that link is quite low.  It's not a broadcast storm or a multicast storm or anything of that sort.  There aren't any spanning-tree instances running on any of the ME3400s so it's not rogue STP traffic that is getting dropped.  None of the MEs are actually seeing any of the errors, only the 7600s.  Any ideas on how I can sniff this out?

- The MEs are all ME3400-24-TS running 12.2(58)EX.
- 76001 is running 15.2(2)S1 76002 is running 15.2(4)S1 with SUP720-3BXL, 6704-3BXL for 10G, 2624-3BXL for 1G.

Thanks in advance for any ideas...