Several of my projects have required multiple Raspberry Pis working in tandem to accomplish an ultimate goal, such as driving multiple independent displays or integrating a device with a dedicated controlling computer. Sometimes the setup had unique hardware (e.g., sensors); other times distance made it easier to use a remote system and WiFi rather than a lot of cabling. Although you can choose from many approaches to distributed technology, here, I'll focus on the Python remote objects (Pyro) library.

My most recent project that fell in this category was a set of scoreboards for my church's Vacation Bible School. I integrated four large LCD TVs into the set design (Figure 1) and dedicated a Raspberry Pi to each one. Also, I wanted the ability to update each team's score in real time from a centralized console. To accomplish this, I wrote the code for the scoreboards and their controller in Python and communicated between the different screens with Pyro [1].

Figure 1: The four monitors as seen from the audio console. Also shown are two screens for Nintendo Entertainment Systems (center) and worship center projector screens (top). The computer barely visible in the bottom left corner accesses the web manager.

Each Pi runs a Raspbian Lite distro and a custom Python script. After flashing the SD card with Raspbian, the only special configuration was to connect to the WiFi network and install the Pyro library. Python is installed by default with Raspbian, and the PyGame library, also a default, provided the graphics. PyGame can drive the Pi framebuffer directly, so the graphical desktop isn't needed.

Elements of a Pyro System

To begin, I'll look at the three parts of a Pyro system and the network individually. Usually, all parts run on separate hardware, but that's not a requirement; they will happily coexist on a single computer for testing, or if it best fits your application. Figure 2 shows how a Pyro network is laid out.

Figure 2: A typical Pyro network and the flow of a request through the system.

Network

For Pyro to work, physical devices must be able to talk across the network. As long as everything is on the same router (either wired or WiFi), you should be in good shape. To check for basic connectivity between devices, use ip a and ping (see the "Checking Network Connectivity" box for more details). If everything is running on a single computer, you're also in good shape.

ping <IP address from above>

Daemons

In Linux, daemons are processes that run in the background to take care of services like printing, checking email, serving web pages, and myriad other tasks. Pyro uses the term "daemon" to describe its workers. A Pyro daemon provides data (class properties) and things it can do (class methods) on the Pyro network. The Python that you write in each daemon makes the provided task happen. Pseudocode looks something like:

@Pyro4.expose
def ringBell():
  bellPin = 17
  GPIO.out ( bellPin , True )
  time.wait ( 1 )
  GPIO.out ( bellPin , False )
  Return "Bell rang for 1 second"

If this looks familiar, you're right! The code to implement your task works just like any other Python code you might write to accomplish a task. The only difference is the decorator on the first line, which lets Python know that it is a Pyro function accessible to the outside world. More on that later.