Aglet

Investigate worker shutdown at safe and mid-step boundaries

Investigation should reveal what the worker promises when it receives a shutdown signal and where restart can safely continue. Use a fixed job with phase markers and stop it at distinct boundaries. Trace durable checkpoint, effect, claim, process exit, requeue, and restart records in one sequence.

Build a useful investigation brief

  1. Prepare phase-controlled stops

    Create a sanitized job with a safe pause before its next effect, a pause after a durable commit, and a pause inside a bounded active step. Record expected ownership and replay behavior. Use a fresh worker and job ID for each stop.

  2. Trace exit and restart ownership

    Align shutdown signal, drain acknowledgement, checkpoint, effect, process exit, requeue, restart claim, and terminal updates. Check whether a new worker can prove the old claim is abandoned. Keep missing process evidence as an explicit reconciliation problem.

  3. Test one drain boundary

    Exercise a bounded drain or checkpoint rule in the same phases and compare item effects, claims, and restart selection. If a mid-step operation must finish, document its safe recovery instead of reporting the entire worker as interrupted without qualification.

What to carry forward

Deliver phase-controlled traces, last durable checkpoint, claim ownership, restart selection, and effect reconciliation. State the smallest shutdown or recovery boundary to review. Preserve uncertainty for an external or in-flight effect that cannot be correlated to the worker exit.

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