Drink blazin' electric death, downtime! Fast, zero-downtime apps for production enviroments.
CoffeeScript HTML CSS JavaScript
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


Rocksteady Build Status

Drink blazin' electric death, downtime!

Fast, zero-downtime apps for production enviroments. Rocksteady runs your node app and keeps it running for you. It can reload your app and browser on file modifications for more productive development and reload on SIGHUP for zero-downtime app upgrades.


$ npm install -g rocksteady


Point rocksteady at your node app and off you go. You can use the rocksteady:

$ rocksteady ./my-app.js

Or require rocksteady into your project and pass it the path directly:



Run rocksteady -h for a complete list of options.

rocksteady server.js [options]

  --port             Specify port to listen on.
  --workers          Number of workers to start.
  --restart-cooldown Seconds to wait before respawning workers that die.
  --force-kill       Seconds to wait before killing unresponsive worker.
  --watch            Watch for and reload server/browser on changes.


rocksteady(serverModule, [options])

Returns a new instance of Master with given options.

Class: rocksteady.Master

Master represents a running app, it is an EventEmitter.

new rocksteady.Master(serverModule, [options])

  • serverModule [String] Should be a path to a module (either JavaScript or CoffeeScript) which exports either a connect/express app or an instance of http.Server
  • options [Object]
    • port [Number] Port to listen on
    • workers [Number] Number of workers to fork
    • forceKillTimeout [Number] Number of seconds to wait before killing an unresponsive worker
    • socketTimeout [Number] Number of ms to wait before socket times out
    • watch [Boolean] Whether or not to watch for changes and reload server/browser.
    • runAs [Object]
      • dropPrivileges [Boolean] Whether to drop privileges if running as root.
      • gid [String] gid to change to
      • uid [String] uid to change to
    • setupMaster [Object] options to pass to cluster
    • logger [Object] logger to use

Event: 'worker:exception'

function (worker, err) { }

Emitted when worker process messages an uncaught exception. err is a deserialized error message.

Event: 'worker:forked'

function (worker) { }

Emitted when worker is forked.

Event: 'worker:killed'

function (worker) { }

Emitted when worker is killed.

Event: 'worker:restarting'

function (worker) { }

Emitted when worker is restarted.

Event: 'worker:listening'

function (worker, address) { }

Emitted when worker is starts listening.

Event: 'reloading'

function () { } Emitted when reload begins.

Event: 'shutdown'

function () { } Emitted when shutdown ends.

Event: 'watch'

function (message) { }

Emmited during development when a worker discovers a file to watch.

  • message [Object]
    • type [String] always 'watch'
    • filename [String] filename to watch, detected by worker.
    • isDirectory [Boolean] whether or not it's a directory