Kubernetes CronJob Guard
Know when the CronJob did not run.
Kubernetes records Jobs and pod events. It does not page you when a schedule silently stops producing useful work. backpulse watches the expected check-in: missed, failed, late, duplicated, or still running.
Commit to the $15/month betaNo agent, operator, or Prometheus stack required.
Failure semantics
The cases a green control plane still misses
The Job never appeared
A suspended CronJob, broken schedule, controller outage, or missed deadline produces no successful Job and no obvious application error.
The pod failed quietly
Image pulls, evictions, OOM kills, and non-zero exits become one failed check instead of another event buried in the cluster.
The run is stuck
A Job that starts but never finishes is reported when its run window expires.
Runs overlap
Duplicate starts expose concurrency-policy mistakes and retries that are doing the work twice.
The schedule drifts
Expected time, grace period, and timezone are explicit, so a late run is different from a missing one.
Failures persist
Alert on the first failure or only after a threshold, without writing PromQL for every scheduled job.
One wrapper
Keep the CronJob you already deploy
Wrap the existing command. backupctl reports start, exit status, and duration. backpulse compares those signals with the schedule and alert threshold you set.
containers:
- name: export
command: ["backupctl", "run", "--url", "https://api.backpulse.io", "--key", "<key>", "--", "/app/export"]30-day validation
Would you pay $15/month for this?
Record a real paid-beta commitment. No charge today; if this wedge reaches the launch gate, we will confirm fit with you before billing.