For test scenario 1, we simulated an unplanned node failure by performing a power-down of Node 1 in Site 1 (ax740xds1N1) through its iDRAC out-of-band management console. The following figures show Node 1, its running VMs, and the volumes owned by Site 1, respectively, in a healthy state in Failover Cluster Manager before initiation of the power-down operation.

Figure 8.   Node 1 (ax740xds1N1) with Up status

Figure 9.   VMs running on ax740xds1N1

Figure 10.    Volumes owned by Site 1 nodes showing Online

A couple of minutes after we powered down ax740xds1N1 using its iDRAC, the node’s status changed to Down, and then to Isolated in Failover Cluster Manager. The VMs’ status also changed to Unmonitored.

Figure 11.    Node 1 status change to Isolated

Figure 12.    VM status change to Unmonitored

As expected, all VMFleet VMs that had been created on ax740xds1N1 in Bangalore automatically restarted on ax740xds1N2. After approximately 5 minutes, all 10 VMFleet VMs had fully restarted on ax740xds1N2.

Note: We configured Bangalore as the preferred site on the ax740xds1N1 volume to ensure that the volume remained active in the Bangalore site. If we did not configure the preferred site, the active volume could be moved to Chennai, and the VMs would be restarted on a cluster node in Chennai.

Figure 13.    VMs restarting on ax740xds1N2

The unplanned node failure simulation had no impact on the volumes owned by the Bangalore nodes or on any resources in the Chennai site.