STP and trunk flapping

[javiero_2]

Hello,

I had a crazy problem occur in a distribution switch. The switch was alternating between active and standby sups causing STP issues. The distribution is fully redundant. Other distribution switches with forwarding trunks through this problem switch never used another path, since the trunk port was flapping. The problem with the switch was a bad slot 1 and secondary sup.

I know this problem of flapping sups is very rare but I needed to try and come up with a preventative measure for the future. There does not seem to be a way to keep a root port from reastablishing itself as a root port when the port has been flapping. Since the faulty switch was the root bridge, it would always become root.

My ideal solution would be to have a mechanism where the switch could somehow take preference of a blocking port during a situation like this.

[skarundi]

I would love that too :)

For now all we have is Unidirectional link detection (UDLD). If the failovers are causing stp loops, UDLD helps prevent the broadcast storm associated with the stp loop by shutting down one of the affected links that may be intermittently sending out BPDUs. Since in this case you had a bad sup, the bad sup uplink port may have been having issues sending out traffic and the adjacent cisco switch may have detected this through the UDLD mechanism and shut down its port.

More details on UDLD is found at

URL: http://www.cisco.com/warp/customer/473/77.html