Although Jobs and CronJob are the special utilities of Pods, the Kubernetes system has different management systems between them and Pods.

For Job, its selector cannot point to an existing pod. It is a bad idea to take a Pod controlled by the deployment/ReplicaSets as a Job. The deployment/ReplicaSets have a desired number of Pods running, which is against Job's ideal situation: Pods should be deleted once they finish their tasks. The Pod in the Deployments/ReplicaSets won't reach the state of end.