EgressIP Failover Dynamics: Normal Behavior and the Impact of Graceful Shutdowns/Reboots

Issue

In a default configuration where egressIPReachabilityTotalTimeout is set to 5 seconds, the ovnkube-cluster-manager actively monitors node health. During a service disruption or a node reboot, the EgressIP must migrate to a healthy node to maintain traffic continuity. This failover process involves health-probe monitoring and Gratuitous ARP (GARP) announcements, which can result in temporary "dual MAC" observations on the network.