Monitor Scheduled Jobs and Application Tasks

Scheduled work is broader than Cron

Cron is one scheduler. Systemd timers, hosted schedulers, CI jobs, platform tasks and application schedulers can all fail before useful work completes. This solution page covers the common operational pattern without duplicating the Cron-specific guide.

Define expected execution and completion

Write down when the job should run, how long normal work takes, what completion means and which evidence is produced. A start event is not completion, and a zero exit code may not prove that useful output exists.

Use the right product concept

Heartbeat Monitor records outbound evidence. Dead Man Monitor is the missing expected check-in model. Dead Hand Monitor is relevant only when active work should advance a meaningful progress marker.

Limitations

HeartbeatHook does not schedule tasks, inspect scheduler logs or validate the business output. Keep scheduler state and application logs as part of the system of record.

Compare Cron-specific monitoring →