[c-nsp] Sup720 software forwarding

[Peter Rathlev]

On Mon, 2013-03-11 at 20:03 +0100, Tóth András wrote:
> MAC learning is fully hardware based on 6500 Sup720, therefore even if
> it's flapping it will not cause packets to be software switched, nor
> CPU usage to increase due to learning.

That's what I've told myself, but something undeniably made the CPU
usage go through the roof and that something is related to MAC flapping.

> Did you rectify the MAC flapping issue? If so, did you see an
> improvement with those flows, i.e. did they disappear from ibc/netdr
> and the CPU usage went down?

Yes. A soon as the MAC flapping stopped (e.g. by shutting one of the
interfaces or adjusting STP costs to make the flapping happen elsewhere)
we only saw "normal" traffic on the IBC and the usual very low CPU load.

> Any traces of warnings in the logs? Any resources exhausted? "sh pla
> ha ca" might reveal it.

No. Actually we were completely stumped by there being nothing relevant
in our logs at all. Until of course we discovered that mac-move
notification hadn't been enabled.

With mac-move notif. enabled we just see the usual messages:

003813: Mar 11 15:00:53.782 CET: %MAC_MOVE-SP-4-NOTIF: Host 0050.568a.5a6c in vlan 208 is flapping between port Te6/2 and port Te6/3
003814: Mar 11 15:00:53.782 CET: %MAC_MOVE-SP-4-NOTIF: Host 0050.568a.5a66 in vlan 208 is flapping between port Te6/2 and port Te6/3
003815: Mar 11 15:00:53.782 CET: %MAC_MOVE-SP-4-NOTIF: Host 0050.5684.47f6 in vlan 282 is flapping between port Te6/3 and port Te6/2
003816: Mar 11 15:00:53.782 CET: %MAC_MOVE-SP-4-NOTIF: Host 0050.5684.47f6 in vlan 282 is flapping between port Te6/3 and port Te6/2

I with there was some way to sess exactly why a packet was punted. We
can recreate this 100% on two switches in production and are currently
getting ready to recreate in the lab. We've tested it on both SXI1 and
SXJ3 and will try other linecards

The powers that be are satisfied with our having found a "solution" but
my curiosity begs me to find out more. Is there any point in trying to
push our partner to take this up with TAC? (After having tested of
course.)

-- 
Peter