March 17, 2013 3k views
I would like to know: in case you have to do planned/scheduled maintenance (reboot, replace hw, etc) on a physical host, do you: a) notify the owners of the droplets? b) shutdown the droplets and reboot - or do you use some save-and-restore-vm-state-to-disk feature of the hypervisor - or do you move the vm's to another physical host?
2 Answers
I've been notified of maintenance by email which is good enough for me. Not sure about the second one.
There are different kinds of maintenances that we sometimes need to do that have differing maintenance windows and allowed time for action.

In each case customers are notified by email regarding the maintenance listing the affected droplets, the reason for the maintenance, what the possible impact may be, and whether or not any action will be taken on the specific virtual servers.

Some maintenances, such as network, do not require the virtual servers to be shutdown in anyway, while others like migrating to a new hypervisor do. On occasion this means that the VM will be powered down and then migrated to another hypervisor - we also do port checking before and after, to make sure that the open listening ports prior to the migration are once again open and listening after.

Have another answer? Share your knowledge.