Aglet

Verify long-job limits below, at, and beyond the deadline

Verification must exercise the exact deadline cases that determine whether a job can be retried safely. Run a below-limit control, a boundary interruption before a commit, and a timeout after a commit or during cleanup. Read durable records and worker ownership after each job settles.

Check whether the outcome improved

  1. Verify a below-limit completion

    Run the fixed job with enough margin to finish its phases and cleanup. Confirm the expected durable output, success state, no timeout marker, and no retry. Record the phase timing as a fixture control, not as a promise for every workload.

  2. Verify timeout before commit

    Pause the job just before the known commit and let the limit fire. Confirm no partial output is presented as complete, cleanup runs safely, and the terminal state gives a clear retry or review path. Reconcile the item ledger with any worker exit record.

  3. Verify timeout after commit

    Cross the deadline after the commit and observe cleanup, status, and retry behavior. Ensure a retry cannot repeat a non-idempotent effect without reconciliation. Check that dependents see the documented state instead of a generic failure or indefinite running label.

What to carry forward

Accept when below-limit, pre-commit timeout, and post-commit timeout fixtures produce the documented effects, cleanup, terminal state, and retry decision. Keep the tested phase and workload visible. Reopen verification if a timeout leaves unknown writes or a dependent cannot distinguish safe retry from 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