Kubernetes CronJobs (not yet published)
Not yet published. A CronRadar controller for Kubernetes CronJobs is built but not released. Its image, Helm chart and repository are not public yet, so nothing on this page can be installed today. Once it is published, this page carries its setup guide.
What it will do
You install the controller once per cluster. It finds every CronJob, in every namespace or the ones you list, and keeps CronRadar in sync with them:
- each CronJob becomes a monitor with its schedule, keyed
<namespace>-<name> - a deleted, suspended or opted-out CronJob's monitor is paused, with its history kept
- each Job a CronJob creates is reported as a start, a completion or a failure, read from the Job's status, with the failing container's exit code
There is no sidecar and no change to your containers. The controller only reads CronJobs, Jobs and pods, and the application's API key is its only required configuration. A CronJob opts out with the annotation cronradar.com/skip: "true".
CronRadar evaluates schedules in the application's time zone. A CronJob whose spec.timeZone differs needs its own application set to that zone, or a schedule rewritten in the application's zone.
Monitor CronJobs now
Until the controller is published, report runs from inside the job: wrap the container's command with the start, complete and fail requests in the full lifecycle example, and give the container CRONRADAR_API_KEY from a Secret. The image needs curl. See cron schedule syntax for the schedules CronRadar accepts.