Kubernetes CronJob generator

Writing a Kubernetes CronJob that behaves

A Kubernetes CronJob is the cluster's version of crontab, but it has more moving parts: it creates a Job on schedule, the Job creates Pods, and each of those layers has its own retry and cleanup settings. The minimal YAML from the documentation runs, but in production the defaults surprise people: overlapping runs when a job takes longer than its interval, finished Pods that pile up for ever, a schedule that fires at the wrong hour because the controller runs in UTC, or a name that is silently too long.

This generator asks for the settings that matter and produces a complete manifest. The schedule is validated with the same Kubernetes cron rules used elsewhere on this site, the time zone is checked against the IANA database, the name is validated against the DNS rules and the 52-character limit, and the container command is split into command and args properly.

Settings worth deciding on purpose

• concurrencyPolicy: Forbid prevents two copies of a batch job from running together.
• startingDeadlineSeconds says how late a run may start (for example after controller downtime) before it counts as missed.
• ttlSecondsAfterFinished removes finished Jobs and Pods automatically, so completed runs do not accumulate.
• restartPolicy: OnFailure restarts the container inside the same Pod; Never creates a new Pod per retry and keeps every attempt's logs.
• Resources: set a memory limit equal to the request so the Pod has predictable behaviour; add a CPU limit only if you have measured throttling.

The generator warns about the common mistakes (unpinned images, overlap allowed, no time zone) rather than silently accepting them. Review the YAML against your cluster version before applying it. Nothing leaves your browser.