That would be a pretty concerning way of handling database failures, but I suspect you might be right.