Aglet

How to verify scheduled work runs once and recovers safely

Verification must show that scheduled work runs at the intended time, handles failure clearly, and recovers without overlapping duplicate effects. Exercise a normal run, dependency failure, delayed execution, missed window, and overlap in a controlled environment, then inspect identifiers, state, and visible outcome.

Check whether the outcome improved

  1. Replay a normal due run

    Use the original schedule identity, timezone, due window, input shape, target data, worker conditions, and expected output. Record calculation, enqueue, claim, start, completion, retry, and final visibility. Compare intended and actual times and confirm one logical run creates the expected effect.

  2. Exercise failure and missed timing

    Cause a controlled dependency or worker failure, delay execution past the due window, and observe retry or missed-run handling. Confirm the status explains what happened, a safe replay exists when intended, and a later normal run is not silently skipped or duplicated.

  3. Check overlap and schedule continuity

    Start two due runs or keep one run active when the next occurrence becomes due. Verify the overlap rule, locks, identifiers, effects, and final state. Continue through the next schedule occurrence and confirm recurring work remains eligible, visible, and correctly separated by schedule identity.

What to carry forward

Classify scheduled work as reliable, regressed, unchanged, or inconclusive with calendar, run, failure, overlap, and final-state evidence attached. It passes only when due runs execute as intended, failures become recoverable or explicit, overlaps are safe, and recurring continuity remains intact.

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