[c-nsp] Loop/Unreachable problem with C6500/SUP720

[Phil Mayers]

On 08/08/2012 08:07 PM, Sebastian Wiesinger wrote:
> * Xu Hu <jstuxuhu0816 at gmail.com> [2012-08-08 18:30]:
>> Are this routes all running in the ospf and bgp at the same time? If yes, it is a normal behaviour.
>
> Hello,
>
> as Gert pointed out, the networks are connected (I also have another
> network, which is a static route (redistributed into OSPF), which is
> showing the same behaviour.

For inspecting this kind of problem I like to use both the "sh ip cef" 
and "mls" variants i.e.

sh mls cef lookup <ip> detail

...followed by:

sh mls cef adjacency entry <xxx>

Where "xxx" is the bit from the first command after the A: i.e.

M(11978  ): E | 1 FFF  0 0 0 0   255.255.255.255
V(11978  ): 8 | 1 1    0 0 0 0   155.198.136.2    (A:82280  ...)

is adjacency 82280.

My guess is you've either

  a. got some random layer2 horrbleness going on (ARP, ICMP redirects) - 
but my guess is you've eliminated that - or

  b. there is an IOS bug or hardware problem causing adjacency or TCAM 
mis-programming.

I assume you have tried reloading the box?

It might be worth trying a full set of GOLD diags (including the 
disruptive ones) on the sup and linecard(s) concerned, just to eliminate 
hardware as a cause.

We don't have SXJ3 installed anywhere, but our boxes on SXJ2 don't 
display these symptoms. Not much help, I know...