Perhaps I'm misunderstanding, but power failure resilience is a system design property. You shouldn't be trying to solve it at the orchestration/provisioning le

Perhaps I'm misunderstanding, but power failure resilience is a system design property. You shouldn't be trying to solve it at the orchestration/provisioning level. Add some caps, have durable memory, change FS, avoid writes to OS partitions, do atomic updates, etc. However, you might find QBee relevant for the other stuff.