ipfailover is triggered when the Virtual controller no longer responds to heartbeats or ARP.
This triggers a protocol by which any remaining nodes in the cluster agree that the node is no longer alive.
One of the nodes in the cluster begins the process of claiming the IP address.
It now has 4 IP addresses its own managment and federation and 2 storage (its own and the failed node)
The ESXi access the simplivty datastore via what appears to it to be an nfs mount,
Once another OVC in the cluster has claimed the faulty sotorage interface then the Datastores are available.
In a 2 node cluster the arbtier service is used to determine if a node is truely faulty or if its a network partition
Once the issue on the faulty node is corrected the ip is released and reclaimed by the origional node.
I am an HPE employee
