Detecting deadlocks in distributed systems

The authors deal with a method for detecting deadlocks in resource sharing for distributed systems. The algorithm is based on sending messages along the edges of the waitfor graph, and is built on a prioritized signaling mechanism which can be implemented on an underlying routing protocol. This algorithm supports multiple resources and multiple outstanding requests. The proposed algorithm avoids the detection of false deadlocks. The algorithm works well even when multiple nodes initiate the deadlock detection algorithm. An informal argument of the proof of correctness of the proposed algorithm is presented. A comparison of this algorithm with other existing distributed deadlock detection algorithms is briefly presented.<<ETX>>