We have just experienced a 5th general loss of service since deploying our
Clean Access in Out-of-Band mode last August. Each failure has lasted over 2
hours and nothing we do helps.  Users who are not already logged in are
unable to do so because the CCA Manager cannot successfully change the
authentication vlan to the access vlan because the switch does not respond
to any SNMP request from the manager.

Is there anyone else who has deployments in OOB mode using stacked Cisco
3750 switches?
If so, have you seen this behavior?

We have found out that only switch stacks with 4 or more elements are
affected, with only one exception.

Cisco TAC currently thinks this is a high CPU issue and that since the SNMP
process has a lower priority, something else may be starving it of
resources.  We have read that any change to the running configuration of the
switch stack (as is done each time a user logs in via Clean Access), the
configuration is replicated to each element in the stack using lots of CPU
so this occurs frequently, but this general SNMP failure occurs at seemingly
random times.

Our Cisco TAC case has been open since late January with no resolution as yet.

If anyone else is experiencing this, or has ideas on what may trigger the
event, please let me know.

Thanks!

-Bill
[log in to unmask]
Network Security Administrator
Colorado State University