Foobartory

An actor model to simulate an automated production chain using Python asyncio.

Here's an example

Quickstart

$ git clone [email protected]:thomasperrot/foobartory.git
$ cd foobartory
$ pip install .

Once you have installed the package, you can run the Foobartory with the following command:

$ foobartory --speed=10 -v

To have details about the available options:

$ foobartory --help
Usage: foobartory [OPTIONS]

Options:
  -s, --speed FLOAT  Fasten the factory by the given factory (1 by default)
  -v, --verbose      Use multiple times to increase verbosity  [x>=0]
  --help             Show this message and exit.

Improvements

I chose to implement this project using Python asyncio. This is a simple and efficient approach, but it could be improved. In particular, a major improvement would be to use a micro-services architectures, where each robot is an independent process. The project would then follow a master slave achitecture, where the master is the Factory, and the slaves are the robots.

• Each robot is an independent process, that would run synchronous Python code, packaged in Docker.
• The factory is also an independent process, running synchronous Python code on the host machine.
• Each robot sends Foo, Bar, FooBar and cash to RabbitMQ independent exchanges. Those four exchanges each have a unique queue, which is listen by all robots.
• When a robot wants to buy a new robot, it sends a message in a RabbitMQ exchange to inform the factory. This exchange is common to all robots, and have a single queue, which is listen only by the factory.
• The Factory is in charge of:
  • starting new robots when receiving a message on the dedicated queue. To do this, the factory can use docker-py package
  • stopping robots once 30 robots have been bought. To do this, the Factory sends a stop message to a dedicated exchange, which has one queue per robot. Every robot must stop when a message is received on that queue.

In a production environment, we could even run the robots in GCP (Cloud Run or Kubernetes). Unfortunately, I did not have enough time to implement it.

Where to go from here

The complete docs is probably the best place to learn about the project.

If you encounter a bug, or want to get in touch, you're always welcome to open a ticket.