A Job-like Pod is suitable for testing your containers, which can be used for unit test or integration test; alternatively, it can be used for batch programs:

1. In the following example, we will write a Job template to check the packages installed in image Ubuntu:

$ cat job-dpkg.yaml
apiVersion: batch/v1
kind: Job
metadata:
  name: package-check
spec:
  template:
    spec:
      containers:
      - name: package-check
        image: ubuntu
        command: ["dpkg-query", "-l"]
      restartPolicy: Never

Note that restart policy for Pods created in a Job should be set to Never or OnFailure, since a Job goes to termination once it is completed successfully.

2. Now, you are ready to create a job using your template:

$ kubectl create -f job-dpkg.yaml
job.batch "package-check" created

3. After creating a job object, it is possible to verify the status of both the Pod and Job:

$ kubectl get jobs
NAME            DESIRED   SUCCESSFUL   AGE
package-check   1         1            26s

4. This result indicates that Job is already done, executed (by SUCCESSFUL = 1) in 26 seconds. In this case, Pod has already disappeared:

$ kubectl get pods
No resources found, use --show-all to see completed objects.

5. As you can see, the kubectl command hints to us that we can use --show-all or -a option to find the completed Pod, as follows:

$ kubectl get pods --show-all
NAME                  READY     STATUS      RESTARTS   AGE
package-check-hmjxj   0/1       Completed   0          3m

Here you go. So why does the Completed Pod object remain? Because you may want to see the result after your program has ended. You will find that a Pod is booting up for handling this task. This Pod is going to be stopped very soon at the end of the process.

6. Use the subcommand kubectl logs to get the result:

$ kubectl logs package-check-hmjxj
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name                     Version                       Architecture Description
+++-========================-=============================-============-======================================================================
ii  adduser                  3.113+nmu3ubuntu4             all          add and remove users and groups
ii  apt                      1.2.24                        amd64        commandline package manager
ii  base-files               9.4ubuntu4.5                  amd64        Debian base system miscellaneous files
ii  base-passwd              3.5.39                        amd64        Debian base system master password and group files
ii  bash                     4.3-14ubuntu1.2               amd64        GNU Bourne Again SHell
.
.
.

7. Please go ahead and check the job package-check using the subcommand kubectl describe; the confirmation for Pod completion and other messages are shown as system information:

$ kubectl describe job package-check
Name:           package-check
Namespace:      default
Selector:       controller-uid=9dfd1857-f5d1-11e7-8233-ae782244bd54
Labels:         controller-uid=9dfd1857-f5d1-11e7-8233-ae782244bd54
                job-name=package-check
Annotations:    <none>
Parallelism:    1
Completions:    1
Start Time:     Tue, 09 Jan 2018 22:43:50 -0800
Pods Statuses:  0 Running / 1 Succeeded / 0 Failed
.
.
.

8. Later, to remove the job you just created, delete it with the name. This also removes the completed Pod as well:

$ kubectl delete jobs package-check
job.batch "package-check" deleted

$ kubectl get pods --show-all
No resources found.