On the monitoring of remote procedure call programs

The author presents the design and implementation of a remote procedure call (RPC)-oriented program monitor which records all events of an RPC-oriented program's execution in the monitor's database. Facilities are provided for programmers to define, choose, and combine events that will be recorded. The monitor has several managing servers which record the events of monitored program parts of their hosts into their local event databases. By using an ordering method, all events of an RPC program can be partially ordered and the event graphs of all program parts can be built. These event graphs are then used to replay the program's execution. Facilities are also provided to define combined events, to control the replay process, and to view the details of the message part of an event.