Testing two failure rates with randomly right censored data and uncensored data