At or around 9am PST one or more API server nodes experienced internal errors causing a complete outage on that node. Some nodes were healthy but traffic that hit a bad node would have failed during the outage. All bad nodes were restored by around 9:45am and system is recovering.