----- Original Message ----- One general question - do all hosts connected to the same UPS ? or each host has a different UPS ? If it's a single UPS than it is easy to detect general outage, in case it's multiple UPSs it a bit harder. I assume it's a standard shutdown command, So the VMs should survive that .. like any other linux process on shutdown. If you want to be on the safe side you can either shutdown or suspend all the VMs (using RestAPI to the engine). If you want to suspend you must take under consideration the fact that suspend takes much more time and you should be able to estimate whether the UPS has enough power till this process is done. The VMs would try to migrate automatically when host becomes non-operational. which is not the case here. Unless your storage or network are out as well. The guests would pause in 2 cases: - you send them pause command - they receive EIO from their storage. After concluding this is a general outage (depends on your env) I would shutdown all VMs enough time before the power runs out from the UPS