Emunet: Design and Implementation | a Debugging Aid for Distributed Programs in Tcp/ip Based Network Emunet: Design and Implementation | a Debugging Aid for Distributed Programs in Tcp/ip Based Network

Most programmers depend on reproducible behavior of a program to locate its bugs. In a sequential program, the execution of the program can easily be repeated and the problematic scenario can be replayed to nd the bugs. When a program is distributed across network, however, the execution order of the program components varies from one run to another due to external factors such as network delay. It is very diicult, often impossible, to repeat the same sequence of execution. This makes the debugging of distributed program challenging. This paper describes a tool called EMUNET that can aid debugging in distributed programming. In EMUNET, the network functions in a distributed program are replaced with the software modules that communicate through memory. The distributed program then becomes a single process with multiple, user-controlled threads of execution. The programmer can compile and test the distributed program as a single uni-processor program when debugging. The network activities are simulated. The debugging programmer can use diierent random seed to try diierence sequences of program execution, each of which is repeatable. In this way, most of the bugs can be found when debugging on a single machine. When this initial debugging process is nished, the programmer can then link the distributed program with actual network library for real execution. No internal program changes are required.