I don't want that a k8s job rerun every day

[jlozano-rh]

I have a k8s Job that should be run only once.

Example:

apiVersion: batch/v1
kind: Job
metadata:
  name: example-job
spec:
  backoffLimit: 0
  completions: 1
  completionMode: Indexed
  parallelism: 1
  # keep the log for 2 days
  ttlSecondsAfterFinished: 172800
  template:
    spec:
      containers:
      - name: example-job
        imagePullPolicy: IfNotPresent
        image: some/image:tag
        command: ["bash", "script.sh"]
        envFrom:
          - configMapRef:
              name: config-urls
        env:
          - secretKeyRef:
              name: config-credentials
        volumeMounts:
          - mountPath: /some/cache/dir
            name: cache-volume
      volumes:
        - name: cache-volume
          emptyDir:
            sizeLimit: 500Mi
      restartPolicy: Never

And this is my Kustomization object:

apiVersion: kustomize.toolkit.fluxcd.io/v1beta2
kind: Kustomization
metadata:
  name: example-job
  namespace: cicd
spec:
  interval: 5m0s
  path: ./kustomize/overlays/dev
  sourceRef:
    kind: GitRepository
    name: example-job
  healthChecks:
    - apiVersion: batch/v1
      kind: Job
      name: example-job
      namespace: cicd
  timeout: 10m
  prune: true
  force: true

But every one or two days, the Job run again (and is not the behavior that I need, because I'm running database migrations on it). What's wrong?

[S-mishina]

@jlozano-rh

But every one or two days, the Job run again (and is not the behavior that I need, because I'm running database migrations on it). What's wrong?

I ran into the same issue and did some digging. One thing that caught my eye was that ttlSecondsAfterFinished is set to 172800, so the Job might be getting cleaned up and recreated after that time.
I’m not completely sure, but this discussion seemed relevant, so sharing it just in case:

ref: #3708