Copy the provided exampleConfig.js into a config.js file, and follow the instructions in the file to fill out your application's Google account credentials. Then just run:
node node2dm.js /path/to/config/file
A sample upstart script is included for easy deployment on Ubuntu.
Messages are pushed to the service using simple UDP datagrams, separated by a colon; here's what our Python application server code looks roughly like:
c2dm_socket = socket.socket(socket.AF_INET, socket.SOCK_DGRAM) encoded_payload = '%s:%s:%s' % ( device_token, collapse_key, simplejson.encode(c2dm_payload) ) c2dm_socket.sendto(encoded_payload, (c2dm_server, c2dm_port))
As you can see from the example, the datagram looks as follows:
We chose UDP because we wanted to minimize latency between our application servers and the node2dm service, and don't have hard delivery requirements for push notifications.
node2dm handles a variety of errors that may be passed back from the C2DM service:
node2dm supports a simple webhook-like callback on a bad token error from Google. To set it up, configure serverCallbackHost / serverCallbackPort / serverCallbackPath / serverCallbackSharedSecret in your config.js, and you should start receive POST requests with this payload:
device_token (the offending device token) message_body (the body of the message that was being attempted; can be useful if you need to match something back up on your end) shared_secret (the serverCallbackSharedSecret, can match up on your end to make sure it's the right server sending you a ping)
node2dm comes with a simple, telnet-able stats function. To use it:
> echo "stats" | nc localhost 8121 uptime: 226 seconds messages_sent: 20583 messages_in_queue: 0 backing_off: false total_errors: 9 rate_limited_tokens: 0 logged_in_to_c2dm: true token_age: 221 memory_rss: 88793088 memory_heapTotal: 11075584 memory_heapUsed: 3658300
The project is BSD-licensed; please fork and submit a pull request for any improvements you'd like to see in node2dm.