What is a Kubernetes CronJob?

Overview

A CronJob is a Kubernetes object that creates Jobs on a schedule. If a job is a task that runs once, a CronJob is the same task running repeatedly at a defined time, much like the cron utility on Linux.

This is useful for recurring tasks: nightly data processing, scheduled backups, periodic report generation, and so on.

Example: CronJob manifest

apiVersion: batch/v1
kind: CronJob
metadata:
  name: nightly-backup
spec:
  schedule: '0 2 * * *'
  jobTemplate:
    spec:
      completions: 1
      parallelism: 1
      template:
        spec:
          containers:
            - name: backup
              image: my-app:1.0
              command: ['python', 'backup.py']
          restartPolicy: Never

Let’s break this YAML down:

• schedule uses standard cron syntax (minute, hour, day of month, month, day of week). The example above runs at 2am every day;
• jobTemplate is the Job template that the CronJob creates on each scheduled run. It has the same structure as a regular job manifest (completions, parallelism, pod template, and so on).

Cron syntax

The schedule field follows the standard five-field cron format:

A few common examples:

• "*/5 * * * *": every 5 minutes;
• "0 9 * * 1-5": at 9am, Monday to Friday;
• "0 0 1 * *": at midnight on the first of every month.

Concurrency and history

CronJobs offer a few settings to control what happens when multiple scheduled runs overlap:

• concurrencyPolicy: decides what to do if a new job is created while the previous one is still running. Allow (default) lets them run in parallel. Forbid skips the new run. Cancel terminates the currently running job before starting the new one;
• successfulJobsHistoryLimit: how many completed jobs to keep in the cluster (default is 3);
• failedJobsHistoryLimit: how many failed jobs to keep (default is 1).

Lateness and missed runs

A CronJob creates jobs based on the configured schedule. However, if the Kubernetes controller manager is down or the cluster is under heavy load, a run might be missed. Kubernetes does not go back and run missed schedules; it simply waits for the next one.