I have a three node cluster connected to a SAN via Fiber Channel. Server 2012. I configured everything and verified the cluster setup etc. Created a test VM, tested live migrations. All good. Today I tested physical failover scenarios by simply pulling
cables. Had a VM on node1, pulled fiber connectivity and it failed to node2. Perfect. Pulled the fiber on node2 it failed to node1, perfect. I placed the VM on node3 and here is where it got interesting. I pulled the fiber cables and no failover. VM stayed
on node3 and was still connected and accessible. I then pulled the NIC cables and it failed over. So how would it stay on node3 when the fiber cables to the san were pulled. node1 and node2 failed as expected. I compared configs on the nodes and all looked
the same. I did not mess with iscsi and the service is not running on any of the nodes. its all fiber. I could even go on node3 and open the clusterstorage / volume1 folder with the fiber cables pulled. No other physical connections to the SAN. I'm just not
sure how this could be possible. Any thoughts??
↧