Our monitor adds another finished housing when it reconnects to an already-high completion. The retained job register names the housing family, not the individual attempt. What should we reconcile before changing the count?
Have the controls and application owners compare what each side retained, including attempt identity, result and whether that result was already committed. If the family register is all you have, it cannot distinguish two attempts on its own. Keep unknown work held for review.
There is an application attempt record, but no matching identifier in the retained PLC data. We cannot assign the old completion confidently. The housing is identified separately for quality review.
And does the replay preserve the original stored history? Starting with an empty file could make a duplicate look like a legitimate first completion. I would keep the unknown retained case distinct from a known result delivered twice.
Those are separate offline cases now. The unknown retains its reason across restart; duplicate-known-result testing is still incomplete. No automatic clearance or resend added to make the warning disappear.