raft

The raft module (built on hashicorp/raft with BoltDB storage) runs a Raft consensus node. You provide an applyLog callback; the replicated-state-machine Apply step dispatches each committed entry to it. Nodes are referenced by an integer handle. Import it with import raft.

Functions

Function	Signature	Description
`start`	`raft.start(opts) → int`	Start a node; returns a handle
`apply`	`raft.apply(h: int, command: string) → bool`	Replicate one log entry through the cluster
`isLeader`	`raft.isLeader(h: int) → bool`	Whether this node is the leader
`leader`	`raft.leader(h: int) → string`	Address of the current leader
`state`	`raft.state(h: int) → string`	Node state (e.g. `"Leader"`, `"Follower"`, `"Candidate"`)
`addPeer`	`raft.addPeer(h: int, id: string, addr: string) → bool`	Add a voter to the cluster
`shutdown`	`raft.shutdown(h: int) → bool`	Stop the node

opts for start:

Key	Description
`nodeId`	Unique node identifier
`addr`	Bind address for Raft transport
`storePath`	Directory for the BoltDB log store
`bootstrap`	`true` to bootstrap a new single-node cluster
`peers`	Initial peer set
`applyLog`	Callback `fn(command)` run for each committed entry

Example

import raft

let node = raft.start({
    nodeId: "n1",
    addr: "127.0.0.1:7000",
    storePath: "./raft-data",
    bootstrap: true,
    applyLog: fn(command) {
        println("applied: " + command)
    },
})

if raft.isLeader(node) {
    raft.apply(node, "set x = 1")
}

Standard library · View as Markdown · llms-full.txt