/lipstick

:lipstick: sticky sessions for Node.js clustering done responsibly

Primary LanguageJavaScriptMIT LicenseMIT

lipstick

Sticky sessions for Node.js clustering done responsibly

what is it?

This module provides sticky sessions for Node.js at the cluster level, routing requests to the underlying workers by producing hashes very quickly.

This module is an alternative to sticky-session, usually most useful when running Socket.IO >= 1.0.0. Remember to also set up the socket.io-redis adapter!

inspiration

This module is inspired on the suggestions outlined by elad in their node-cluster-socket.io repository. All credit goes to them.

I built upon the idea by adding support for IPv6, and attempted to streamline the implementation as much as possible for the lipstick consumers so that only minimal changes to your application are needed.

requirements

  • Node.js >= 0.12.x for pauseOnConnect

install

npm install lipstick --save

usage

To use lipstick, your master has to communicate with their workers effectively.

master

Here's a ready to use production-grade cluster.js file. It will listen on port PORT as defined in your environment variables. It will use app.js as your worker process and spawn os.cpus().length workers, or 2 of them, whichever is bigger. It will also route messages to the workers based on a hash of their IP address, applying the stickiness.

require('lipstick')();

master api

The API for your cluster.js module is detailed below.

lipstick(appfile?, options?, done?)

The appfile defaults to ./app.js and will be used as the worker process. Options are detailed below.

Option Description
port The port your application listens on
workers The amount of workers your cluster should spawn

workers

You'll have to make a slight modification in your app.js workers. Just change the following line:

app.listen(port, cb);

To the lipstick equivalent shown below:

require('lipstick').listen(app, port, cb);

This will allow lipstick to patch your worker processes. node app will work as usual.

license

MIT