Kubernetes Jobs (one off tasks)

In the last post I create a simple application to be used within a schedule, i.e. a CronJob in Kubernetes.

We can also create one off tasks (or a Job) which might be used to migrations or some batch processing. We’ve going to use everything from the previous post to build, containerize and push our imager to a container registry. The only change is to use the supplied job.yaml file, listed below

apiVersion: batch/v1
kind: Job
metadata:
  name: one-time-job
  namespace: dev
spec:
  template:
    spec:
      containers:
      - name: one-time-jobb
        image: putridparrotreg/putridparrot/crj:1.0.0
      restartPolicy: Never

Running the following “kubectl get jobs -n dev” results in something like this

NAME           STATUS     COMPLETIONS   DURATION   AGE
one-time-job   Complete   1/1           5s         41s

and if we get the pods using “kubectl get jobs -n dev” we get something like

NAME           STATUS     COMPLETIONS   DURATION   AGE
one-time-job   Complete   1/1           5s         83s

and if we check the pods with “kubectl get pods -n dev” we’ll see something like this

NAME                 READY   STATUS      RESTARTS   AGE
one-time-job-h5dvf   0/1     Completed   0          3m23s

and ofcourse we can see the logs of this run via “kubectl logs one-time-job-h5dvf -n dev” and we get our application output, i.e. the date/time it was run

Current date and time: 2025-08-17 15:39:53.114962479 +00:00

You’ll note that the pod remained in the cluster, this allowed us to view the logs etc. and it’s down to the developer/devops to delete the job and pod unless…

We can actually set up automate deletion of the pod using the “ttlSecondsAfterFinished” option in the yaml file, i.e. 

apiVersion: batch/v1
kind: Job
metadata:
  name: one-time-job
  namespace: dev
spec:
  ttlSecondsAfterFinished: 300  # Deletes Job and its Pods 5 minutes after completion
  template:
    spec:
      containers:
      - name: one-time-jobb
        image: putridparrotreg/putridparrot/crj:1.0.0
      restartPolicy: Never

We also have the option of “activeDeadlineSeconds”, this does not delete or clean up anything but it can be used in the “spec:” section like “ttlSecondsAfterFinished” to denote that the job will be killed off it not finished. So for example

apiVersion: batch/v1
kind: Job
metadata:
  name: one-time-job
  namespace: dev
spec:
  ttlSecondsAfterFinished: 300  # Deletes Job and its Pods 5 minutes after completion
  activeDeadlineSeconds: 600 # Job will be killed even if not finished, in 10 minutes
  template:
    spec:
      containers:
      - name: one-time-jobb
        image: putridparrotreg/putridparrot/crj:1.0.0
      restartPolicy: Never