[c-nsp] IOS XR on ASR9001: Some LDP on Interfaces stuck in xmit

Richard Hartmann richih.mailinglist at gmail.com
Wed Mar 26 07:31:41 EDT 2014


Dear all,


I am having issues bringing up LDP on several interfaces on an ASR9001
with IOS XR 4.3.4. All other machines I am referring to in this email
and in the rest of our backbone are happily talking with each other.

In the context of this setup two ASR9001 and two ME3600x are relevant.
All four machines use different IOS versions, so I will designate them
as:


Machines:
* ASR-4.3.4
* ASR-4.3.0
* ME-15.3(2)S1
* ME-15.3(1)S

Relevant interconnections:
ASR-4.3.4 Te0/0/2/1 <> Te0/2 ME-15.3(2)S1
ASR-4.3.4 Te0/0/2/2 <> Te0/1 ME-15.3(1)S
ASR-4.3.4 Te0/0/2/3 <> Te0/0/2/3 ASR-4.3.0



Now the problem is that TenGigE0/0/2/2 and TenGigE0/0/2/3 are stuck in
xmit, while TenGigE0/0/2/1 goes into xmit/recv as it should:

RP/0/RSP0/CPU0:ASR-4.3.4#sh mpls ldp discovery
Local LDP Identifier: ASR-4.3.4:0
Discovery Sources:
  Interfaces:
    TenGigE0/0/2/1 : xmit/recv
      LDP Id: ME-15.3(2)S1:0, Transport address: ME-15.3(2)S1
          Hold time: 15 sec (local:15 sec, peer:15 sec)

    TenGigE0/0/2/2 : xmit

    TenGigE0/0/2/3 : xmit

  Targeted Hellos:
    ASR-4.3.4 -> C6500 (active/passive), xmit/recv
      LDP Id: C6500:0
          Hold time: 90 sec (local:90 sec, peer:90 sec)

    ASR-4.3.4 -> C6500 (active/passive), xmit/recv
      LDP Id: C6500:0
          Hold time: 90 sec (local:90 sec, peer:90 sec)

    ASR-4.3.4 -> ME-15.3(1)S (active/passive), xmit/recv
      LDP Id: ME-15.3(1)S:0
          Hold time: 90 sec (local:90 sec, peer:90 sec)

    ASR-4.3.4 -> different ME with 15.3(1)S (active/passive), xmit/recv
      LDP Id: different ME with 15.3(1)S:0
          Hold time: 90 sec (local:90 sec, peer:90 sec)

    ASR-4.3.4 -> C6500 (active/passive), xmit/recv
      LDP Id: C6500:0
          Hold time: 90 sec (local:90 sec, peer:90 sec)

    ASR-4.3.4 -> ASR-4.3.0 (active/passive), xmit/recv
      LDP Id: ASR-4.3.0:0
          Hold time: 90 sec (local:90 sec, peer:90 sec)

RP/0/RSP0/CPU0:ASR-4.3.4#sh run
<snip>
ipv4 access-list mpls-ldp-advertisement
 10 permit ipv4 Loopbacks any
 20 permit ipv4 Interconnects any
 30 deny ipv4 any any
</snip>
<snap>
mpls oam
!
mpls ldp
 discovery targeted-hello accept
 label
  allocate for mpls-ldp-advertisement
 !
 interface TenGigE0/0/2/1
  discovery transport-address interface
 !
 interface TenGigE0/0/2/2
  discovery transport-address interface
 !
 interface TenGigE0/0/2/3
  discovery transport-address interface
 !
!
</snap>

mpls-ldp-advertisement is the same on all machines and allows the
relevant loopback and interface IPs.
Adding/removing `discovery transport-address interface` does not have
any effect.
The other sides of the interconnects are happily talking LDP among
each other. All interconnects are configured to accept MPLS

Other than an upgrade of ASR-4.3.0 and ME-15.3(1)S I am out of ideas,
but I would prefer to do this _after_ I got the above to work if at
all possible...


Thanks,
Richard


More information about the cisco-nsp mailing list