chaoshead the TL/DR is this:
• Mediator appliance replaces Pure1 when ActiveCluster is used in Dark Sites
• Depending on how you have your stretched pod configured
◦ If the Mediator is "lost" first, the pod will perform a pre-election. This pre-election will determine which array will continue to make the pod/volumes available if the 2 arrays have an interruption in replication.
◦ If the arrays lose their replication link to each other, they will both "race to the mediator" and whoever wins will present the pod/volumes from the array that communicates with the mediator first.
◦ You can set array preference in the stretched pod which will (in both cases above) fail the pod/volumes to the preferred array.
◦ Keep in mind that you can have multiple pods with the preference set individually per pod.
