[nsp] Catalyst errdisable

[Łukasz Bromirski]

Hi,

 > We added Cat3550-2 to provide some resiliency for the market Ethernet
 > segment, Then we got the dreaded "%ETHCNTR-3-LOOP_BACK_DETECTED:"  which
 > shutdown the .1q trunk between the switches.  I have experienced this
 > behavior before, but now we are getting MALLOCFAIL messages in the logs
 > of both switches (yes, the reported time is off).

Are You getting first MALLOC-FAILs then LOOP_BACK_DETECTED or the other
way? If the MALLOC is first, it causes switch to behave somewhat
mysteriously and I would discard the next "errors", as they are propably
caused by process memory corruption. If the Loopback is first, then the
Mallocs, You have propably some mysterious problem with EtherChannel.
Can You post this trunk config for both sides?

Either way, Your malloc problems seem very similar to ones of my
client (I've post to this list few days ago looking for a solution).
His case was resolved by just upgrading the IOS to 12.1(20)EA1 from
12.1(12c)EA1, but I haven't found any bug in bug toolkit  similar to
observed problems resolved since 12.1(12c)EA1.

Best regards,
-- 
Łukasz Bromirski                             lbromirski:mr0vka,eu,org