Our UR5e verification application crashed while a request was in flight. The local record still says pending, and startup currently sends all pending work again.
I have paused that automatic resend in the test environment. What should the recovery screen show when the original execution might have continued?
Tell us it might already have run. Our old screen just said try again, which sounds like permission to repeat it. Show the job identity and who can check its state.
That is useful. The application currently shows the fixture name but no request identity. Two requests for that fixture would look identical to the operator.
Does the receiving application retain an ID or result you can look up after restart? I'd check that before designing a screen around information you won't have.
Yes, and don't ask the operator to confirm something only the software could know. We can inspect a fixture under the proper procedure; we can't read a lost request out of thin air.
The receiver keeps a local execution log, but no lookup service. I am asking the programmer to establish what survives restart and how a request could be matched before we offer any automatic recovery.
For now the screen can say it needs investigation. That's less convenient than a retry button, but at least the next shift won't unknowingly send the same work twice.
Draft screen now shows the request ID, submission uncertainty and the named support route. Never-submitted records remain separate. Receiver reconciliation is still being designed.
Try that handover with someone who wasn't in this discussion. Ask them what they'd do next. You'll find out pretty quickly whether the words actually help.