If a system fails only 1% of the time in the field but never in the lab, how do you approach finding the root cause?
