[c-nsp] ip cef / ospf redundancy testing

Adam Greene maillist at webjogger.net
Tue Dec 30 16:23:36 EST 2008


Maybe simplifying the question will help.

When I do a "sh ip cef exact-route <src1> <dest1>" on the 3750, it shows me 
that traffic should flow over link A. However, in actuality, it flows over 
link B. When I do a "sh ip cef exact-route <src1> <dest2>" it shows me that 
traffic should flow over link B. However, it actually flows over link A.

When I use a different <src> address but the same <dest> addresses, the "sh 
ip cef exact-route" command gives me accurate results.

I don't have any policy routing enabled. I'm trying to figure out why 
sometimes the "sh ip cef exact-route" command would produce accurate output 
and why sometimes it wouldn't.

Stumped.

(p.s. thanks for the off-list reply I received)

----- Original Message ----- 
From: "Adam Greene" <maillist at webjogger.net>
To: <cisco-nsp at puck.nether.net>
Sent: Tuesday, December 30, 2008 2:14 PM
Subject: [c-nsp] ip cef / ospf redundancy testing


> Maybe someone could point me in the right direction ...
>
> I'm doing some redundancy testing in the lab ...
>
> I basically have two 3750's linked together redundantly via OSPF. IP CEF 
> is enabled. Both switches are running 12.2(25)SEE2. Switch B is receiving 
> redundant default routes from switch A.
>
> According to "sh ip cef exact-route" issued on switch B, traffic from 
> <src1B> to <dest1A> and from <src2B> to <dest2A> should be all be 
> travelling over a single link, leaving the other one empty of traffic. 
> However, what I am actually seeing is that traffic from <src1B> to 
> <dest1A>  is flowing over one link, and from <src2B> to <dest2A> is 
> travelling over the other.
>
> "sh ip cef detail" confirms that per-destination load balancing should be 
> occurring, and when I run a single stream of traffic, indeed round-robin 
> per-packet load-sharing is not occurring.
>
> Why would the traffic be following a different path than what the "sh ip 
> cef exact-route" command suggests it should?
>
> Maybe I need to provide more detail for anyone to be able to opine ...
>
> Thanks,
> adam
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>
> 




More information about the cisco-nsp mailing list