Check whether the outcome improved
Establish the first-attempt ledger
Run a fixture with valid and invalid rows and inspect accepted, rejected, and unresolved outcomes. Compare the outcome report with destination state. Confirm the retry interface explains which category it will select and does not imply that a failed overall status means every source record remains unwritten.
Exercise correction and repeated retry
Correct the rejected row through the supported path and review the exact next-attempt selection. Retry again where the contract permits it. Check that already accepted work receives the documented treatment and that a changed file order does not silently attach the correction to a different original source record.
Reconcile final row identity
Compare every final record with the intended source and inspect updates as well as creations. Confirm remaining failures stay visible and accepted rows are not mislabeled as pending. Record any interruption case separately when its durable outcome cannot be established through the available controlled workflow or outcome report.
What to carry forward
Accept when each attempt’s scope is clear and final per-row state matches the documented retry contract. Keep unresolved first-attempt outcomes outside a blanket pass. A successful second job is insufficient evidence unless already accepted records, corrected failures, and any still-pending rows have all been accounted for explicitly.
Technical background: Google Cloud 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