Aglet

Verify heartbeats for short, long, and blocked job phases

Verification must cover the phase that originally went quiet and the phases that provide a control. Run short work, a slow known phase, and a bounded blocked call. Confirm the monitor reports liveness or review according to evidence, while the job’s terminal outcome remains independent.

Check whether the outcome improved

  1. Verify a short phase

    Run a job whose progress and heartbeat updates are frequent enough for the monitor. Confirm timestamps, worker ownership, and terminal success or failure agree. This control catches a general sender or observer problem before the long-phase case is interpreted.

  2. Verify the long phase

    Hold the job in the documented slow step and observe the heartbeat or phase-progress signal across the monitor interval. Confirm it remains classified as active or reviewable without claiming completion. Read the worker and durable state after the phase finishes.

  3. Verify a blocked dependency

    Pause a job on a known dependency and check the liveness status, wait reason, and eventual release or timeout. Confirm a blocked job is not treated as healthy progress merely because a heartbeat exists, and not retried blindly because progress is quiet.

What to carry forward

Accept when short, slow, and blocked phases produce distinct, evidence-backed liveness and terminal states, with no premature duplicate intervention. Keep the monitor interval and phase assumptions recorded. Reopen verification if a heartbeat masks a blocked job or a quiet phase is marked dead without review.

Technical background: Celery documentation.

Keep the decision with the work.

Use a Work Item in Aglet to record the problem, the evidence you have, and the next decision. Add an owner and priority, then keep updates in the discussion so the next person can follow the reasoning.

Create an account See the product workflow