From 9:19 to 9:28 we experienced some errors and timeouts on 10 to 15% of requests from Asia. There was a temporary saturation of connections following a failed update of a threshold during a platform update. The problem has been fixed and a solution is being implemented to prevent it from happening again.