Cron jobs

Use cronJobs to run scheduled work on the same container model as deployments and jobs. Each key becomes a CronJob, named after the key and inheriting every global default (image.*, imagePullPolicy, configure.*, and the rest).

Minimal example

A bare key produces a working cron job that runs @daily with a single container named after the key:

cronJobs:
  backup:       # runs the default image @daily

More often you set a schedule and a command:

cronJobs:
  example-cronjob:
    schedule: "1 * * * *"
    command: ["echo", "Hello world!"]

For a release named test, that renders CronJob/test-cronjob-example-cronjob. See Naming for how the cronjob- prefix and resource name are built.

Safe by default

Every cron job is generated to be safe unless you say otherwise:

Field	Default	Meaning
schedule	@daily	Standard cron syntax or an @ macro.
concurrencyPolicy	Forbid	A run is skipped if the previous one is still going.
backoffLimit	0	A failed job is not retried.
restartPolicy	Never	A failed pod is not restarted in place.

Override any of them per cron job:

cronJobs:
  report:
    schedule: "@weekly"
    concurrencyPolicy: Replace    # cancel the running job and start the new one
    backoffLimit: 3               # retry a failed run up to three times
    command: ["/app/bin", "generate-report"]

Scheduling and job options

Beyond the defaults above, the CronJob and Job spec fields below are available, rendered only when you set them. The common case is timeZone: without it a schedule runs in the kube-controller-manager's zone (UTC in practice), so 30 2 * * * is not 02:30 where you work.

cronJobs:
  report:
    schedule: "30 2 * * *"
    timeZone: Europe/Amsterdam      # 02:30 local, follows DST
    successfulJobsHistoryLimit: 3
    ttlSecondsAfterFinished: 86400  # clean up finished jobs after a day
    command: ["/app/bin", "generate-report"]

Field	Level	Meaning
timeZone	CronJob	IANA zone the schedule runs in (e.g. Europe/Amsterdam).
startingDeadlineSeconds	CronJob	Deadline to start a missed run before it counts as failed.
suspend	CronJob	Pause scheduling without deleting the cron job.
successfulJobsHistoryLimit	CronJob	How many completed jobs to keep.
failedJobsHistoryLimit	CronJob	How many failed jobs to keep.
parallelism	Job	Pods that may run at once.
completions	Job	Successful pods needed for the job to finish.
completionMode	Job	NonIndexed (default) or Indexed.
activeDeadlineSeconds	Job	Wall-clock limit before the job is stopped.
backoffLimitPerIndex	Job	Retry limit per index; requires completionMode: Indexed.
maxFailedIndexes	Job	Failed indexes tolerated before the job fails (Indexed jobs).
podFailurePolicy	Job	Rules matching pod failures, rendered verbatim.
successPolicy	Job	When an Indexed job counts as succeeded, rendered verbatim; requires completionMode: Indexed.
ttlSecondsAfterFinished	Job	Seconds to keep a finished job before it is cleaned up.
podReplacementPolicy	Job	When a failed pod is replaced.

Job-level fields render under jobTemplate.spec. Numeric fields render whenever the key is present, including at zero, so successfulJobsHistoryLimit: 0 keeps no completed jobs rather than falling back to a default.

The Kubernetes API rejects a Job that sets both backoffLimit and backoffLimitPerIndex. When you set backoffLimitPerIndex, the chart omits backoffLimit, including its usual 0 default, so the two never render together.

Containers and pod settings

A cron job carries the same pod and container fields as a deployment. The bare-key form is shorthand for a single container; declare containers explicitly when you need more than one, or to set container-level fields:

cronJobs:
  sync:
    schedule: "*/15 * * * *"
    containers:
      sync:
        image: myorg/sync:latest
        command: ["/app/bin", "sync"]
        configure:
          secrets: true     # mount the release's Secret into this container

Pod-level fields like volumes, volumeMounts, nodeSelector, and configure.* cascade to the containers exactly as they do for deployments. See Conventions for bare keys, cascading, and opting out with [] / {}.

Disabling a cron job

Set enabled: false to keep a cron job in values but skip rendering it, which is handy for per-environment overrides:

cronJobs:
  nightly-import:
    enabled: false    # exists in values, renders nothing
    schedule: "@daily"
    command: ["/app/bin", "import"]

One-off jobs

For work that runs once (migrations, install hooks) rather than on a schedule, use jobs instead. It follows the same container model. See Conventions for the bare-key form and the Helm hook annotations jobs supports.