socketio-pg 0.13

Websocket server using PostgreSQL as a message transport. Uses SocketIO, Greenlet, Flask.

SocketIO PostgreSQL PubSub Websocket Server

What Is This?

This is a websocket server that uses PostgreSQL as its message transport system.

If you have an application that already uses postgres, you can easily send and receive asynchronous events via your existing database. Check out the postgres documentation on LISTEN/NOTIFY for more details.

How Do I Use This?

Server

Configuration

First configure a database connection string in DATABASE_URL. You can create a local.cfg:

# local.cfg
SQLALCHEMY_DATABASE_URI = 'postgresql:///mydatabase
DEBUG = True

Or set DATABASE_URL in your environment. If you use Heroku and have a database attached, this will be set already.

Prerequisites

Note: python 3.6 or higher is required. pip install -r requirements.txt

Run

• Debug server mode: DEBUG=1 python socketio_pg/server.py
• In gunicorn (production): gunicorn --worker-class eventlet -w 1 socketio_pg.server:app
• If using Heroku, a Procfile is already set up for you.

Client

On the client side, you simply connect a socket.io client to begin sending and receiving events. There's a simple demo HTML page at socketio_pg/static/test.html that you can access from the dev server at http://localhost:3030/static/test.html.

Why Use This?

If your application already uses PostgreSQL, you can start sending and receiving asynchronous events right away. It makes an excellent transport for messages (keep them small though, under 8000 bytes), and you can simply issue queries to do it. No additional infrastructure needed, besides this websocket server. If you aren't using PostgreSQL, maybe you should be.

One neat trick is to set up triggers that emit NOTIFY queries when rows on certain tables are inserted or updated. This allows messages to be delivered to clients notifying them of updates without any application code at all. Some demos and slides from a talk can be found here.

In Action

How Is This Built?

This server uses the following technologies:

• Flask - python web microframework
• psycopg2 - postgresql driver
• eventlet - lightweight green threading library
• greenlet - app threads in python
• socket.io - client and server layer on top of websockets to handle namespacing, reconnection, and basic session management