Rafting Trip - August 3-7, 2020

This was my code from David Beazley's amazing Rafting Trip training course, where we tried to implement the raft distributed consensus algorithm in a week.

I didn't finish, but I got fairly far. Code here for my archive.

• transaction log class with accept/reject logic for append requests
• vaguely ports & adapters architecture, with raft protocol classes as core domain
• optional persistent log storage adapter
• log replication leader -> follower, with backtracking algorithm
• heartbeats + time events
• elections, including timeouts, conversion to candidate, voting, promotion to leader
• RaftNet abstraction for transport, with fake version for tests
• run_server() and clock_tick() abstractions for running the algo over time
• ability to unit test / simulate interaction of multiple nodes in-memory
• end-to-end and integration tests
• client interactions <-- this is the big missing piece, which might have caused me to rethink the architecture. may require some intermediary piece in between Server objects / run_server() and clients.
• cluster membership changes