This repo contains my implementations of mit 6.584 and notes.
All tests have been passed.
Summary:
- Lab1 implements a map-reduce system, including basic crash handling for workers.
- There is a coordinator running as a RPC server and tasks scheduler.
- Splits input files to map tasks
- Provide AllocateTask, which allocate a idle task to a worker
- Provide SubmitTask, which receive a task finished by a worker
- Create reduce tasks according to completed map tasks
- Watching one task to be finished in 10 seconds. If timeout, assume the worker has crashed, and reschedule the task.
- There are workers running to fetch tasks, solve tasks, and submit tasks.
Notes:
- notes/lab1.md