It would be good to post-facto expose some indication (in the Activity Feed, or a logfile or alert) that an unplanned VM migration occurred at the cloud provider level. This would allow customers to identify and distinguish cloud provider hardware issues from cluster issues arising from other causes.