Raft is a consensus algorithm. It allows distributed systems to keep a shared and managed state (https://raft.github.io/). Setting up a Raft system is complex in many ways – for one, you need consensus for an election to occur and succeed. This can be difficult to bootstrap when you're working with multiple nodes and can be difficult to get started. It is possible to run a basic cluster on a single node/leader. However, if you want redundancy, at least three nodes are needed to prevent data loss in the case of a single node failure. This concept is known as quorum, where you must maintain (n/2)+1 available nodes to ensure new logs can be committed to the Raft cluster. Basically, if you can maintain quorum, the cluster remains healthy and usable.

This recipe implements a basic in-memory Raft cluster, constructs a state machine that can transition between certain allowed states, and connects the distributed state machine to a web handler that can trigger the transition. This can be useful when you're implementing the base finite state machine interface that Raft requires, or when testing. This recipe uses https://github.com/hashicorp/raft for the base Raft implementation.