Kubernetes Tutorial · Workloads · Lesson 12 of 26

Jobs and CronJobs

Run batch work to completion, retry on failure, and schedule recurring tasks.

Create a free account to save this page, take notes and track your progress. Sign in

Jobs

A Job runs pods until a number of them complete successfully. Useful fields: backoffLimit (retries), completions and parallelism, activeDeadlineSeconds, and ttlSecondsAfterFinished to clean up.

CronJobs

A CronJob creates a Job on a schedule ("0 3 * * *", with an optional timeZone). Set concurrencyPolicy: Forbid when runs must not overlap, and make jobs idempotent: a job can run twice (retries, missed schedules).

An indexed Job and a CronJob

kubectl create namespace demo-jobs > /dev/null
trap 'kubectl delete namespace demo-jobs --wait=false > /dev/null' EXIT
cat > job.yaml <<'EOF'
apiVersion: batch/v1
kind: Job
metadata:
  name: report
spec:
  completions: 3
  parallelism: 3
  backoffLimit: 2
  template:
    spec:
      restartPolicy: Never
      containers:
        - name: worker
          image: busybox:1.37
          command: ["sh", "-c", "echo processed shard $JOB_COMPLETION_INDEX"]
  completionMode: Indexed
EOF
kubectl -n demo-jobs apply -f job.yaml > /dev/null
kubectl -n demo-jobs wait --for=condition=Complete job/report --timeout=120s > /dev/null
kubectl -n demo-jobs get job report -o custom-columns=SUCCEEDED:.status.succeeded,FAILED:.status.failed
for i in 0 1 2; do kubectl -n demo-jobs logs "$(kubectl -n demo-jobs get pods -l batch.kubernetes.io/job-completion-index=$i -o name)"; done
kubectl -n demo-jobs create cronjob nightly --image=busybox:1.37 --schedule="0 3 * * *" -- echo cleanup > /dev/null
kubectl -n demo-jobs get cronjob nightly -o custom-columns=SCHEDULE:.spec.schedule,SUSPENDED:.spec.suspend,CONCURRENCY:.spec.concurrencyPolicy

kubectl create namespace demo-jobs > /dev/null
trap 'kubectl delete namespace demo-jobs --wait=false > /dev/null' EXIT
cat > job.yaml <<'EOF'
apiVersion: batch/v1
kind: Job
metadata:
  name: report
spec:
  completions: 3
  parallelism: 3
  backoffLimit: 2
  template:
    spec:
      restartPolicy: Never
      containers:
        - name: worker
          image: busybox:1.37
          command: ["sh", "-c", "echo processed shard $JOB_COMPLETION_INDEX"]
  completionMode: Indexed
EOF
kubectl -n demo-jobs apply -f job.yaml > /dev/null
kubectl -n demo-jobs wait --for=condition=Complete job/report --timeout=120s > /dev/null
kubectl -n demo-jobs get job report -o custom-columns=SUCCEEDED:.status.succeeded,FAILED:.status.failed
for i in 0 1 2; do kubectl -n demo-jobs logs "$(kubectl -n demo-jobs get pods -l batch.kubernetes.io/job-completion-index=$i -o name)"; done
kubectl -n demo-jobs create cronjob nightly --image=busybox:1.37 --schedule="0 3 * * *" -- echo cleanup > /dev/null
kubectl -n demo-jobs get cronjob nightly -o custom-columns=SCHEDULE:.spec.schedule,SUSPENDED:.spec.suspend,CONCURRENCY:.spec.concurrencyPolicy

Output

SUCCEEDED   FAILED
3           <none>
processed shard 0
processed shard 1
processed shard 2
SCHEDULE    SUSPENDED   CONCURRENCY
0 3 * * *   false       Allow

Run this on your computer — see “Get Started” for setup.

See Jobs and CronJob.

Test yourself

Why must CronJob workloads be idempotent?

CronJobs run in parallel always
A scheduled job can run more than once (retries, missed schedules), so repeating it must be safe
Kubernetes requires it syntactically
To save CPU

Show answer

Answer: A scheduled job can run more than once (retries, missed schedules), so repeating it must be safe

Design jobs so a second run causes no harm.

Was this page helpful?