Our UR5e sample-check application saved pending, sent the request and then crashed before storing an answer. Startup wants to send it again. I have stopped that automatic retry in the test version, because pending covers both unsent work and work whose outcome we lost. I am unsure what the recovery screen should offer while that difference is unresolved.
Replying to John_Blair
Show the saved attempt identity and its unknown outcome without offering an automatic repeat. What evidence survives on the PLC and application sides to reconcile that attempt? The agreed recovery route needs a responsible person and must survive another restart without quietly returning to retry.
7 points