Thanks for the feedback all. I turned off IP redirects and the CPU usage on the MSFC (which was a result of the traffic being punted to the MSFC as originally expected) dropped down to zero and there are no more input drops or traffic showing up as switched by the MSFC.
A TAC case was opened and it turned out to be (most likely) because the hosts (actually other routers and firewalls) were disregarding the IP redirects the MSFC in question was sending, which forced all of the traffic to continue to be switched by the MSFC. Apparently anything that requires an IP redirect be sent will be punted to the MSFC. If the hosts ignore those IP redirects, all packets (which is what made me think it was maybe flow based) that host sends are punted to the MSFC. What confused me is why the traffic was getting punted to the MSFC; I read references to two factors that could cause this. 1) IP redirects and 2) IP cef adjancency of the same interface for the routes in question. I didn't think it was the IP redirects because that just didn't make sense to me, however that was the fix. And this will fix a similar situation I've seen on other MSFC's as well, that just werent as obvious, so this is good. The problem started when the MSFC became a router on a stick... thus it began sending out IP redirects. I realize router on a stick isnt the best design principle but it works and sometimes it's more effort to avoid then it's worth. _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
