Cluster Storage Validation May Fail After Power Outage

If two or more cluster nodes are simultaneously powered on or rebooted prior to the test, it may result in the hardware TickCount to be the same.  An identical TickCount across multiple machines may cause the random number generator used to create the key during the test to generate the same persistent reservation key on different nodes.  A TickCount on one node within 16 milliseconds of the TickCount on another node could result in a duplicate key being generated.  This issue is most likely to occur after a power outage to a datacenter and all servers simultaneously regain power and boot.  This issue is unique to the Cluster Validation test and does not impact production clustered disks, as the Physical Disk resource uses a different mechanism to generate persistent reservation keys.

To resolve this issue, reboot all of the nodes in the cluster in a staggered fashion, waiting at least 1 second between POST on each cluster node. (A 1 second delay between POST on the cluster nodes will avoid the 16 millisecond window that can lead to duplicate keys).  Then run the Cluster Validation tool again. If this Validate Disk Arbitration test continues to fail after a staggered power on or reboot, please contact your storage vendor, as the failure indicates that the storage is not handling arbitration properly. This will be addressed in a future version of Windows.