Public | Automated Build

Last pushed: 2 years ago
Short Description
Short description is empty for this repo.
Full Description


Simple real-time messaging in web applications.

Quick start

pip install centrifuge

Go to http://localhost:8000 - this is an administrative interface of Centrifuge node you just started. More about installation and configuration in documentation.

You can also run Centrifuge in a Docker container by running:

docker build -t centrifuge
docker run -it -p 8000:8000 centrifuge


In a few words: clients (users of your web application/site) connect to Centrifuge from browser,
after connecting clients subscribe on channels. Every message which was published into that
channel will be delivered to all clients which are currently subscribed on that channel.

To connect to Centrifuge from browser pure Websockets
or SockJS library can be used. So it works in both
modern and old browsers (IE 7 supported). Centrifuge has javascript client with simple API.

Backend is built on top of Tornado - fast and mature
asynchronous web server which can handle thousands of simultaneous connections.

Centrifuge scales using Redis PUB/SUB capabilities.
Single full-featured instance of Centrifuge run by default without extra dependency
on Redis.

Centrifuge comes with administrative web interface to manage project/namespace
structure and monitor important messages in real-time.

Persistent data (projects, namespaces) by default stored in SQLite database.
When running Centrifuge instance processes on different machines MongoDB
or PostgreSQL backends can be used instead of SQLite. There is an option
to hard-code all these settings in JSON file and go without any dependency on database.

There are tons of examples in internet about how to add real-time events on your site. But very few
of them provide complete, scalable, full-featured, ready to deploy solution. Centrifuge aims to be
such a solution with simplicity in mind.

Main features

  • Asynchronous backend on top of Tornado
  • SockJS and pure Websockets connection endpoints
  • Simple javascript client
  • Presence information, message history, join/leave events for channels
  • Web interface to manage your projects
  • Flexible channel settings via namespaces
  • Language agnostic - you can go with Centrifuge even if your site built in Perl, PHP, Ruby etc.
  • Easily integrates with existing web site.

To get more information:

Various packages and tools related to Centrifuge located in Centrifugal
organization here on Github.

Similar projects / alternatives:

Basic usage from browser

var centrifuge = new Centrifuge({
    url: 'http://localhost:8000/connection',  // Centrifuge SockJS connection endpoint
    project: 'PROJECT_ID', // project ID from Centrifuge admin interface
    user: 'USER_ID', // your application user ID (can be empty for anonymous access)
    timestamp: '1395086390', // current UNIX timestamp (number of seconds as string)
    token: 'TOKEN', // HMAC token based on project's secret key, project ID, user ID and timestamp

centrifuge.on('connect', function() {


    var subscription = centrifuge.subscribe('django', function(message) {
        // message from channel received

    subscription.on('ready', function(){
        subscription.presence(function(message) {
            // information about who connected to channel at moment received
        subscription.history(function(message) {
            // information about last messages sent into channel received
        subscription.on('join', function(message) {
            // someone connected to channel
        subscription.on('leave', function(message) {
            // someone disconnected from channel


centrifuge.on('disconnect', function(){


For more information about javascript client API see documentation chapter

Admin web interface


IMPORTANT! At moment tests require Redis running and clear database on every running. Be aware of this!

pip install tox

or just

python test


Pull requests are welcome! But, please, follow next principles:

  • keep things simple
  • pep8 friendly
  • python 2.6, 2.7, 3.3 and 3.4 compatible



Docker Pull Command
Source Repository