Can you add an additional option for the Beaconing (ARP monitor) that would implement a dead-node timer? Or number of retries before a fail-over is performed? This would help mitigate some false positives during heavy switch load or packet loss during high traffic.