This project runs on a Raspberry Pi. Pi-System-RPC-Service connects to a RabbitMQ server and listens for messages. When a message is received on the appropriate queue, some system data is gathered (only CPU temperature for now) and returned to the client that sent the message. This architecture makes it simple for any process, running anywhere (like Holly), to access this data as needed.

Using Fabric there is an installation task included in this project's fabfile. With the Raspberry Pi connected to the same network as your computer, run the following command:

fab raspberry_pi install

The task will prompt you for these values:

• Raspberry Pi: - The hostname of the Raspberry Pi, for example: red-dwarf. Given the example, the Fabric script will attempt to ssh into red-dwarf.local.
• Loggly token: - The token from your Loggly account. The service logs data using Loggly which makes it easier to remotely monitor.
• Loggly domain: - The domain from your Loggly account.
• Rabbit URL: - The connection URL for the RabbitMQ server. If you don't feel like running your own, check out CloudAMPQ.

The install process will add an Upstart script that will handle starting/stopping the service when the Raspberry Pi starts up or shuts down.

To manually stop it:

sudo service system-rpc stop

To manually start it:

sudo service system-rpc start

Any script or program can request data from this service provided:

• It has the same Rabbit URL value used during installation and can connect to the RabbitMQ server.
• It sends messages to the correct queue (system.get in this project).

There are plenty of JavaScript client libraries for RabbitMQ. This example uses Jackrabbit.

var jackrabbit = require( "jackrabbit" );

// Use an environment variable for RABBIT_URL
var broker = jackrabbit( process.env.RABBIT_URL, 1 );

var ready = function () {
    // Send a message to request the system data
    broker.publish( "system.get", {}, function ( err, data ) {
        if ( err ) {
            // Do something with the error
            console.log( err );
        }
        // Do something with the sensor data
        console.log( data );
        process.exit();
    } );
};

var create = function () {
  broker.create( "system.get", { prefetch: 5 }, ready );
};

broker.once( "connected", create );