[c-nsp] Monitoring CPU punted traffic

Peter Rathlev peter at rathlev.dk
Wed Sep 10 05:16:42 EDT 2008


On Tue, 2008-09-09 at 16:33 -0400, Jeff Fitzwater wrote:
> I am running 720CXL with SXH code and am trying to monitor the punted  
> traffic to the RP so that I can confirm what actually gets punted to
> it.
> 
> It appears to show packets but not positive I have configured it  
> correctly.   Has anyone else used this tool?

What kind of traffic do you see? What do you expect to see, i.e. what's
missing?

I don't know the SXH way (sounds fancy though) but there's the "good
old" RP-inband/SP-inband way, described here:

http://www.cisco.com/en/US/products/hw/switches/ps708/products_tech_note
09186a00804916e0.shtml#span_inband
http://tinyurl.com/5hx7pb

Docs says you "have" to use the new semantics on SXH though, just like
you did. You might not be able to use the above on SXH, or it might be
unsupported.

> This is how I configured the SPAN port....
> 
> (config) monitor ses 1 local   (This puts me in config-mon-local mode)
> 
> (config-mon-local) destination interface gi13/17  (THIS PORT HAS  
> TCPDUMP HOST ATTACHED)
> 
> (config-mon-local) source cpu rp tx  (As stated in doc traffic is from 
> viewpoint of ASIC, so I used TX assuming transmitted traffic to RP
> CPU.)
> 
> (config-mon-local)no shutdown  (This is needed to turn on monitor)
> 
> (config-mon-local) exit  (Must exit in order for no shutdown to take  
> effect)

Small point: According to the docs, you should rather use "local-tx"
type than "local" when it's just for TX. It's apparently a matter
resources and not functionality though.

Regards,
Peter




More information about the cisco-nsp mailing list