[nsp] Catalyst errdisable

[Dale W. Carder]

We've seen MALLOC's on our network, especially during an STP root 
change.

During that time, and perhaps related to bug CSCed65818, some of our 
3550's
go MALLOC crazy, with exact same I/O pool message posted earlier.  
Normally,
STP (R-PVST+ in our case) works itself out, the 3550's calm down, and 
business
is normal.

However, be warned.

Sometimes on a busy enough link the 3550 can get into a state when 
during a
time of MALLOC's, the switch can become unresponsive.  It doesn't 
crash, but
it effectively locks up, management interface goes dead, front panel 
buttons do
nothing, gives up on switching, turns into a hub, and creates network 
loops that
further exacerbate the problem.  Think positive feedback loop.  Then 
you get to
do truck rolls to physically reboot switches all over town.

Cisco's response ranges from "this is not a problem" to "just try some 
rate-limiting".

Welcome to the 3550 MALLOC club.

As for the loop problem, check out bugs CSCdy88935 and CSCdt82690. 

Dale

-----------------------------------------------
Dale W. Carder
Network Engineer
University of Wisconsin at Madison