[nsp] RPF problem with ICMP unreachables

[Hank Nussbacher]

On Sun, 24 Nov 2002, Stephen J. Wilcox wrote:

> I would assume that to take the source address of the exiting interface would be
> the most sensible approach.. as this interface is participating in the switching
> path towards the destination (of the icmp) you would assume it has a valid ip
> address and is better than a random address which may come from a private net
> running on the same router. No need to keep state either.
> 
> I thought this was the current behaviour?

So did I.  I'll try to get the customer version and config.

-Hank

> 
> Steve
>