Jobs are a familiar concept in many systems. Jobs in K8S are pods that execute a task and runs to completion. Unlike, Deployments, Statefulsets, Replicasets that are meant to run continuously (e.g. they are restarted when they fail, and keeps on running to no well-defined end), Jobs perform a task to completion then end. In this case, end means completed successfully. If the Job fails, it can be configured to restart. Therefore, Jobs are for those tasks that need to complete successfully, and be cleaned up once that objective is met.
#deploy the sample job. This job just prints a message to stdin 30 times then exits.
kubectl create -f job.yml
kubectl get jobs/batch-job
#this shows that we desire 4 successful completions
#and currently there are 0 successful completions
NAME DESIRED SUCCESSFUL AGE
batch-job 4 0 3mYou can also schedule jobs to run at a certain time. See CronJobs at K8S.io.