Let’s take the following Kubernetes Cronjob

apiVersion: batch/v1
kind: CronJob
metadata:
  name: sleep-cronjob
spec:
  schedule: "*/5 * * * *"
  jobTemplate:
    spec:
      backoffLimit: 0 # Avoid two executions
      template:
        spec:
          containers:
          - name: sleep
            image: busybox
            command: ["sh", "-c", "sleep 120"] # 2 minutes
          restartPolicy: Never
          terminationGracePeriodSeconds: 300 # 5 minutes

It runs a job every 5 minutes, and the pod sleeps for 2 minutes.

If during the sleep I execute the kubectl delete pod <POD_NAME> for a graceful termination, it will work, it will wait for the specified terminationGracePeriodSeconds and in this case, it will allow the pod to finish its job successfully.

But, the Kubernetes Job will be marked as 1 Failed with the reason BackoffLimitExceeded.

I have tried the Job configuration activeDeadlineSeconds and I’m still getting the error.

Am I missing a configuration or something for the CronJob/Job to prevent this?