clu

clu

cluster without ster... name might change

A cluster manager inspired by cluster. Still under development!

Features:

• built in CLI (optional)
• zero downtime restarts
• restarts workers one after another (to prevent performance hit)
• add or remove workers on the fly
• uses the node cluster API

Setup

1. npm install --save clu
2. Create a server.js that starts your app.

var clu = require("clu");

clu.createCluster({
	exec: "./app.js",
	workers: 2,
	silent: false,
	silentWorkers: true,
	cli: true
});
// short: clu.createCluster("./app.js");

clu.use(clu.repl());

Start your cluster with node server start or node server & (if you have the cli disabled)

Commands

node server --help will output all available commands

These only work if you have the 'cli' option enabled.

repl

Only works if you clu.use(clu.repl()) Throws you in a repl.

start

Will start the server and throw you back into your terminal.

stop

Will stop all workers and the master.

restart

Respawns all workers one after another.

restart-master

Will restart the master process. (This will cause a downtime!).

scaleup x

Start x number of workers

scaledown x

Stop x number of workers Use --force to kill them

API

Methods

clu.createCluster(options)

Creates a new cluster.

Options

• exec - the file to execute (required)
• workers - Number of workers, will default to the number of cores
• silent - Boolean: Makes the master & workers silent (default: false)
• silentWorkers - Boolean: If true workers output won't be displayed (default: true)
• cli - Boolean: Enables the CLI. (default: true)

clu.restart(cb)

Respawns all workers one after one and calls the callback after all workers have been respawned.

clu.restartMaster(cb)

Restarts the master process. Calls the callback when all workers have disconnected.

clu.stop(cb)

Stops the master process and all workers. Callback gets called when all workers disconnect.

clu.stopWorkers(cb)

Stops the all workers but not the master process. Callback gets called when all workers disconnect.

clu.scaleUp(num, cb)

Spawns x new workers. Will cb after all new workers are listening.

clu.scaleDown(num, cb)

Stops x amount of workers. Will cb after all workers have disconnected. Will throw if you stop more workers than available or you try to stop all (use clu.stop() or clu.stopWorkers() for that).

clu.workers(cb)

Calls back with an array of workers. Returns if no callback is given.

clu.status()

Returns some status data. Example:

{ 
	workers: { 
		total: 2,
		active: 2,
		pending: 0,
		averageUptime: 22.2365 
	},
	master: {
		uptime: 22.4090,
		memoryUsage: {
			rss: 17698816,
			heapTotal: 12376832,
			heapUsed: 4889800 
		}
	}
}

Plugins

Plugins can be used like this:

var cluAwesome = require("clu-awesome");
clu.use(cluAwesome());
clu.use(clu.repl());

Built in:

• clu.repl() - a repl interface

Own:

• clu-dnode - dnode interface for clu

3rd Party: contact me if you create any :)

Licence

MIT