Containers and virtual machines are not relocated to other nodes in the cluster with Parallels Cloud Server nodes and Parallels Cloud Storage if one of the nodes is rebooted or powered off gracefully.
Rebooting the server or even shutting it down does not cause virtual machines and containers to move. By design, graceful reboot or clean shutdown is treated as a planned operation and such operation does not initiate relocation of clustered resources to other nodes.
Fail over of VMs and containers happens only after unexpected loss of functionality (like a crash).
High Availability feature can be tested by imitating a crash of one of the nodes, for example, using the following command which turns off the networking on the node (this command should be used from the physical console of the server or within the session of
~# service network stop; sleep 1200; service network start
(Here, the network on the node will be down for 20 minutes and the service
shaman will consider the node as crashed and thus it will relocate the resources.)
For more information about tuning shaman configuration, check this article or
man shaman.conf command on any of the nodes.