A tote stops at a communication point. The shift logs it, the ticket says “happens sometimes”, and by the time development looks at it the situation is long gone. Nobody can say which telegram arrived in which order, or whether an acknowledgement was missing at all.
So the search starts from scratch every time: read logs, guess, rebuild by hand, hope it happens again. Meanwhile the fault reappears on the next shift.
TeleTrace records the real telegram traffic and replays it against a test system. The fault happens again, on demand, in a system where you are allowed to debug.