A distributed program is presented that ensures delivery of a message to the functioning processors in a computer network, despite the fact that processors may fail at any time. All processor failures are assumedd to be detected and to result in halting the offending processor. A reliable communications network is assumed.