My guess is that there is a procedural issue that will turn up when the situation is monitored closely. We have had no similar reports from the field, and won't be able to do much until the problem is repeatable.