[c-nsp] ME3600X <==> ME3600X IPv6 IS-IS Issue - Update!

[Mark Tinka]

So for the benefit of the archives, this turned out to be an 
issue related to the BFD for IPv6 command being present in 
the code, but the actual feature not being there.

When BFD for IPv6 is enabled on an ME3600X port facing a 
non-ME3600X device (including a device from Cisco), IS-IS 
for IPv6 functions well. However, when BFD for IPv6 is 
enabled between ME3600X devices, IS-IS for IPv6 fails.

We have BFD for IPv6 enabled on links where the ME3600X is 
adjacent to Cisco XR 12000's, Cisco CRS as well as Juniper 
MX routers. In this case, IS-IS for IPv6 works fine.

For the links between two ME3600X devices, enabling BFD for 
IPv6 prevents IS-IS for IPv6 from running. This is quite 
curious, but is the issue.

If anyone else hits this issue, the solution is to disable 
BFD for IPv6 when adjacent devices are ME3600X's.

Mark.

On Wednesday, November 05, 2014 08:56:24 AM Mark Tinka 
wrote:
> Hi all.
> 
> I have one that has baffled me these past days and I'm
> hoping some fresh eyes on the list can find something I
> missed.
> 
> I'm going to open a case with Cisco, but given how
> painful that is, I thought I'd start here first.
> 
> I'll summarize the situation based on the troubleshooting
> done so far.
> 
> Topology is:
> 
> 	XR12000 <==> ME3600X-1 <==> ME3600X-2
> 
> Code is 15.4(2)S on the ME3600X's.
> 
> There is a link pending delivery between the XR12000 and
> ME3600X-2. I'd like to fix this issue before this link is
> in place, for reasons you will see below.
> 
> In short, IS-IS routing for IPv4 and IPv6 is fine between
> XR12000 and ME3600X-1. However, only IS-IS routing for
> IPv4 works fine between both ME3600X's.
> 
> The IS-IS IPv6 database on the ME3600X-2 shows that all
> the IS-IS-learned IPv6 routes via ME3600X-1 are in the
> local database, but for some reason, they do not end up
> in the IS- IS IPv6 RIB:
> 
> me3600x-2#sh isis ipv6 rib
> IS-IS IPv6 process 1, local RIB
> me3600x-2#
> 
> The adjacencies between both ME3600X's are good. This is
> the one from ME3600X-2:
> 
> me3600x-2#sh clns is-neighbors detail
> Tag 1:
> System Id      Interface   State  Type Priority  Circuit
> Id Format
> me3600x-1   Te0/2       Up     L2   0         00
> Phase V
>   Area Address(es): 49.0004
>   IP Address(es):  xxx.xx.x.53*
>   IPv6 Address(es): FE80::2E1:6DFF:FEA1:76DB
>   Uptime: 00:40:02
>   NSF capable
>   BFD enabled: (MTID:0, ipv4) (MTID:2, ipv6)
>   Topology: IPv4, IPv6
>   Interface name: TenGigabitEthernet0/2
> me3600x-2#
> 
> And this is the one from ME3600X-1 (you will see the one
> adjacency to the XR12000 on te0/1 and the other to
> ME3600X-2 on te0/2):
> 
> me3600x-1#sh clns is-neighbors detail
> Tag 1:
> System Id      Interface   State  Type Priority  Circuit
> Id Format
> xr12000-Te0/1       Up     L2   0         00
> Phase V
>   Area Address(es): 49.0001
>   IP Address(es):  xxx.xx.x.49*
>   IPv6 Address(es): FE80::1E17:D3FF:FE12:9078
>   Uptime: 1d20h
>   NSF capable
>   BFD enabled: (MTID:0, ipv4) (MTID:2, ipv6)
>   Topology: IPv4, IPv6
>   Interface name: TenGigabitEthernet0/1
> me3600x-2   Te0/2       Up     L2   0         00
> Phase V
>   Area Address(es): 49.0004
>   IP Address(es):  xxx.xx.x.54*
>   IPv6 Address(es): FE80::2E1:6DFF:FEA1:855B
>   Uptime: 00:41:33
>   NSF capable
>   BFD enabled: (MTID:0, ipv4) (MTID:2, ipv6)
>   Topology: IPv4, IPv6
>   Interface name: TenGigabitEthernet0/2
> me3600x-1#
> 
> Everything looks good, IS-IS IPv6 database has all the
> right data, but nothing makes it into the IS-IS IPv6 RIB
> on ME3600X-2. Consequentely, the FIB does not get
> updated, ending up with no useful IPv6 forwarding on
> ME3600X-2.
> 
> Suffice it to say that IPv6 pings between both ME3600X's
> work fine (directly connected routing).
> 
> Also, the IS-IS IPv6 RIB on ME3600X-1 looks good, with
> routing learned from the XR12000. So IPv6 forwarding is
> fine on this unit.
> 
> Has anyone seen anything like this before?
> 
> I have a truckload of other ME3600X's with good IS-IS
> IPv6 routing, but these have successful adjacencies and
> IPv6 routing/forwarding with Juniper MX480's, Cisco
> CRS's or Cisco XR12000's. This is the first one we're
> doing with adjacencies between like-for-like ME3600X's.
> 
> All help appreciated.
> 
> Mark.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
URL: <https://puck.nether.net/pipermail/cisco-nsp/attachments/20141207/2aa9592e/attachment.sig>