[c-nsp] Strange CDP Observation

Anthony Ryan anthony.ryan at manchester.ac.uk
Mon Nov 30 05:53:27 EST 2009


On 28 Nov 2009, at 22:42, Felix Nkansah wrote:

> One of my field engineers at a remote site has reported a problem to  
> me, one
> which I find rather strange but yet to confirm for myself.
>
> A Catalyst 2960 switch at the remote site is linked to a switch at a  
> Central
> site via radio (bridged).
>
> However, besides the fact that connectivity tests between the two  
> sites are
> failing, his observation is that when he issues the "SHOW CDP NEI"  
> command
> on the remote C2960 switch, he sees the switch itself appearing as its
> neighbor rather than the other switch at the central location which is
> typically the output he expected.
>
> I have personally never seen a Cisco device showing itself as it's own
> neighbor in a 'show cdp neighbor' output.
>
> What do you know to be the cause of this problem?

We used to see symptoms like fairly regularly that with certain  
provider's telco circuits when they went wrong. Usually it meant that  
a line card somewhere within the telco circuit had gone into a soft- 
loop mode. This sort of soft-loop is a deliberate and useful feature  
when building circuits and testing each leg of the circuit at  
installation time as it means they can test each section of the link  
without needing someone to physically put a loop of cable at the far  
end.

But if the soft-loop feature had somehow randomly switched on then it  
meant there was a problem with our circuit. It could happen due to a  
software bug, or a hardware fault...or maybe human error if an  
operator at the telco remotely applied a loop on the wrong device when  
trying to commision a new circuit somewhere else. Sometimes it would  
go away by restarting the telco equipment. Sometimes apparently it  
needed a linecard change ( in the telco kit not in our Cisco kit. )  
And maybe sometimes somebody at the telco quietly removed the soft- 
loop without admitting they created it by mistake.

If there is a loop in place you might be able to detect/prove it even  
more by forcing some traffic out out and watching the packets counts  
in/out increase by the same amounts as each packets you transmit  
arrives back to you on your receive.

I dont know the nature of your radio circuit but it could be a problem  
with a soft-loop within that equipment perhaps. Do you control all the  
hardware for that or do you lease the circuit from someone ?

Anthony Ryan







More information about the cisco-nsp mailing list