Background jobs

TL;DR. Every @Scheduled job gets RED metrics, an in-flight gauge, and a health check that flips DOWN when a job hasn't succeeded for too long. Stuck nightly jobs page on-call instead of going unnoticed.

@Scheduled jobs are the silent killers of Spring services. They run on a private thread, log to the same place as everything else, and the first sign that one has been failing for a week is when a downstream system breaks because the nightly reconciliation never ran.

Pulse gives every job RED metrics, an in-flight gauge, and a health indicator that flips DOWN when a job hasn't succeeded inside a configurable grace period — so the on-call sees a stuck job before its absence breaks something else.

What you get

sum by (job) (rate(pulse_jobs_executions_total{outcome="failure"}[1h]))

A failing job lights up here long before its silence is felt downstream.

The /actuator/health/jobs indicator flips DOWN when any job hasn't succeeded inside its grace period — so Kubernetes (or whatever consumes the health probe) can surface it as a real failure, not a buried log line.

Turn it on

Nothing. Every @Scheduled job is observed automatically. ShedLock-managed jobs work too — the decorator wraps the Runnable before the scheduler sees it.

To tune the grace period (default 1 hour):

pulse:
  jobs:
    failure-grace-period: 6h

What it adds

Metric	Type	Tags
pulse.jobs.executions	Counter	job, outcome (success / failure)
pulse.jobs.duration	Timer	job, outcome
pulse.jobs.in_flight	Gauge	job (overrun detector)

Plus the /actuator/health/jobs indicator.

When to skip it

If you already monitor your scheduled jobs through Quartz JMX or a job-runner-specific dashboard:

pulse:
  jobs:
    enabled: false

---

Source: io.github.arun0009.pulse.jobs · Status: Stable since 1.0.0