Some of the periodic jobs should actually be running continuously instead of trying to approximate runtime for a cron-like execution.

Create a mw-daemon (feel free to bikeshed the name) MW-on-K8s deployment that:

• Takes a yaml list of maintenance scripts and parameters similar to periodic jobs
• Creates Deployments and single-replica ReplicaSets with restartPolicy: Always for these maintenance jobs

This would continually relaunch the maintenance job on Pod termination, error or otherwise.

Things to consider:

• Alerting
• scap deployments