1. Proposed title of this feature request

Move Machine-config-daemon to a systemd service, and implement MC rollbacks.

2. What is the nature and description of the request?

Currently, if a customer rolls out an MC that results in a node becoming not-ready. or any other edge case that results in the machine-config-daemon pod not starting, there is no way to recover the node without manual intervention on the node its self.

If the machine-config-daemon was deployed as a systemd service then it could recover the node and roll back to the last known good config without manual intervention. or roll forward to the next-good if the customer reverted the MC  that caused any issues.

3. Why does the customer need this? (List the business requirements here)

We have had many cases where invalid configs had to be manually recovered, the worst is probably bad proxy/registry mirroring configs that result in the kublet not starting.

4. List any affected packages or components.

Machine-config-opeartor