On Distributed Systems Debugging

Software engineers have to face many problems when creating, testing and debugging their applications. One of the most challenging problems today is closely associated with the modification of distributed systems. Even a small modification of such a system can considerably change its behavior. This paper explores methods which play an important role in the research on debugging parallel and distributed systems. We will study the detection of global state predicates and the use of temporal logic operators. Finally, on-line and off-line approaches to the problem of debugging a distributed system will be discussed. These include real-time monitoring and deterministic replay. 1