Interactive debugging in a distributed computational

TL;DR: The notion of probable simulation is defined, and the debugging facility is shown to achieve it, which simulates a particular computation of the program it is being used to debug that is likely to occur.

Abstract: This thesis describes an implementation of a facility for interactively debugging distributed programs. These distributed programs consist of groups of cooperating processes concurrently executing on an arbitrarily extensive network of processors. The facility allows the user to monitor and control, at his leisure, the interprocess communications that occur through message passing while execution of the distributed program proceeds. It presents the user with the ability to simulate transmission errors and delays, to alter and create packets, and to precisely control the pattern of such communications. The facility serves as a tool for the detection of lurking bugs, those errors, peculiar to parallel processing, which may or may not appear during the course of any particular execution. The facility possesses a high degree of transparency towards the program being debugged. That is, it has a minimal effect on the events that define the execution of that program. Transparency is a desirable property for any debugger to possess. To achieve such transparency, the processes of the distributed program are made to execute in a logical time environment, reading logical, rather than physical, clocks. We show that the facility obeys a clock condition, with which any logical time system must comply in order to be correct. We also show that the facility actually simulates the program it is being used to debug. Finally, we show that the facility simulates a particular computation of the program that is likely to occur. The notion of probable simulation is defined, and our debugging facility is shown to achieve it.