The overall outcome (and the student's experience and stress) is ultimately a staff issue.

The professor should have been on call, and should have been able to get the student back into the questions, or, failing that, to offer another sitting. I have done both of these, among other remedies, and setting student expectations ahead of time is crucial for avoiding massively stressful problems (stuff like "If something goes wrong, stop and email me. I can add lost your time back and get you into the test again, no matter how long it takes to sort out. You won't lose marks from this." and actually follow through). Software inevitably fails sometimes, and it is just a tool. Staff need to make their own decisions.