Check whether the outcome improved
Verify an in-window read
Complete the fixed job and read its output before expiry. Confirm the job ID, output revision, content, metadata, and success state agree. Record the response as the availability control and ensure a fresh request can find the same result.
Verify the expiry boundary
Read at or just around the recorded expiry and compare the result with retention policy. Check whether the response is available, marked expiring, or explicitly unavailable. Keep clock precision and any cache behavior visible so the boundary is interpretable.
Verify post-expiry recovery
Read after expiry and follow the documented rebuild or empty-state path. Confirm the reader cannot see a stale success label, and that any regenerated output has a new identity or revision. Reconcile downstream consumers before accepting a rebuild as equivalent.
What to carry forward
Accept when in-window, boundary, and post-expiry reads follow the stated retention and recovery contract, with job and output identity visible. Reopen verification if output disappears early, readers see stale success, or rebuild silently repeats effects or changes the result without notice.
Technical background: Rails guides.
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