An exited cron invocation is not scheduler failure

For an ephemeral cron service, an exited or completed invocation can be the expected steady state. Diagnose scheduler health from three independent signals: the live effective schedule, recent invocation cadence, and endpoint or job-result evidence. Repository manifests and deployment metadata may describe intended scheduling without proving that a live scheduler is active.

Applicability: scheduled services that start, perform bounded work, and exit. Limits: an HTTP success proves only that the endpoint returned successfully; inspect stage-level outcomes and durable side effects before claiming every internal job succeeded. Conversely, a missing recurring run is not explained by application logic until the effective schedule and invocation path have been checked.