Our FR5 inspection bench retains job registers through an outage. When the monitor returns it can count the retained completion as new. I need to reconcile it with the job already recorded, without resending inspection or clearing the retained evidence. Where should the reconnect path begin?
Begin with a snapshot of the retained identity and result, plus the monitor's durable history. Compare them before changing totals. Reconnecting establishes that you can observe state again; it does not establish that another inspection occurred.
Keep an offline example of that snapshot before changing the handler. Deliver it again after a simulated reconnect and inspect the report total as well as the record list. They may not use the same counting path.
I've saved it. Retained state has the job label, but that label was reused and our local history doesn't distinguish the two attempts. A duplicate list check alone won't establish which one it belongs to.
Do not let a new deduplication rule create certainty the old data lacks. If the record cannot be assigned, retain it as unresolved and require the appropriate review rather than attach it to the active job.
Quality owns disposition separately. Controls is checking whether any further attempt identity exists in the retained exchange; nothing has been found yet. I've stopped this record being counted automatically in the offline draft.
Make that hold survive another restart in the test. If reopening turns unresolved into ordinary pending work, the first careful screen has not prevented the same mistaken association later.
Future attempts need a durable identity shared across the relevant exchange and monitor history. That can prevent repetition of the ambiguity; it cannot retrospectively identify the older completion simply because the new code has a better field.
Offline reopening preserves the unresolved record without recounting or resubmitting. Controls still hasn't established an older attempt identifier, so I cannot assign this completion. Future identity design is a separate change request.
The report draft excludes it from accepted totals and shows an unresolved entry. Quality's separate disposition has not returned, and the real interface review is still open.
Keep the source snapshot with that entry for the reviewer. The honest result may remain unassignable, but the next person should be able to see why without repeating your reconnect experiment on the cell.