An (N-1)-resilient algorithm for distributed termination detection

The paper presents a fault tolerant termination detection algorithm based on a previous fault sensitive scheme by Dijkstra and Scholten. The proposed algorithm can tolerate any number of crash failures. It runs as efficiently as its nonfault tolerant predecessor if no process actually fails during the computation, and otherwise incurs only a small amount of cost for each actual failure. It is assumed that the underlying communication network provides such services as reliable end to end communication, failure detection, and fail flush