MAC move causing a 5-minute unicats flood?

[milan.kulik]

Hi, does anybody know what happens exactly when a switch has noticed a particular MAC source address received on a new port while the original port did not come Down? The switch will send a "MAC address flapping between ports ...." error message to the Syslog, of course. But as the original port is still Up, there is no STP TCN triggered. And the original port MAC table entry is not cleared immediately, I believe. So until the MAC address aging timer (5 minutes) expires for the original port, the same MAC entry is present on two ports? And the switch is forwarding frames with that destination MAC address to both ports? Or even flooding them to all ports using the "unknown unicast" flood procedure? (This seems to be observed by my colleague.) I'll probably need to build a lab to test, but possibly somebody knows the answer already? Thanks, Milan

[Aaron Harrison]

Hi

The MAC address would be learned immediately and removed from the original port. No STP change/TCN would be caused by a proper 'edge' port changing state.

Unicast flooding would only occur in multiple switch networks, where the traffic destined TO that MAC traverses switches that are not in the path FROM that MAC address (i.e. where the switches have not learned the MAC address port location, as they learn only based on source address). A common example of this is with dual HRSP setups: http://www.cisco.com/c/en/us/support/docs/switches/catalyst-6000-series-switches/23563-143.html

But you should definitely build a lab, that's how you learn :-)

Aaron