Lesson  in  Kubelings — Learn Kubernetes the Rustlings Way

CronJob Pileup: fix the concurrencyPolicy

A every-minute CronJob runs longer than a minute, so overlapping runs stack up and exhaust resources. Configure the CronJob so a new run never starts while the previous one is still going.

The situation

The report CronJob in kubelings is scheduled every minute, but each run takes ~90 seconds. With concurrencyPolicy: Allow, Kubernetes starts a fresh run on schedule even though the previous one is still running — so active Jobs pile up and pods accumulate until the namespace runs out of resources.

Your task

Configure report so a new run does not start while the previous run is still active (set concurrencyPolicy to Forbid, or Replace if you'd rather kill the old run and start fresh).

kubectl -n kubelings get cronjob report -o yaml | grep concurrencyPolicy
kubectl -n kubelings get jobs
Hint
kubectl -n kubelings patch cronjob report --type=merge \
  -p '{"spec":{"concurrencyPolicy":"Forbid"}}'
Solution

Root cause

report runs every minute but each run lasts ~90s, and concurrencyPolicy: Allow lets a new run start on schedule regardless of in-flight runs. Overlapping Jobs accumulate until resources are exhausted.

Fix

kubectl -n kubelings patch cronjob report --type=merge \
  -p '{"spec":{"concurrencyPolicy":"Forbid"}}'
  • Forbid — skip the new run if the previous one is still active.
  • Replace — cancel the running one and start the new one instead.

Verify

kubectl -n kubelings get cronjob report \
  -o jsonpath='{.spec.concurrencyPolicy}{"\n"}'
kubectl -n kubelings get jobs   # runs no longer stack up
Previous lesson
The Job That Never Finishes
Next lesson
Autoscale a Deployment with an HPA (1 → 5)
Start lesson