Our instances started working again, so seems to be fixed