igondiu / inter_process_communication

Inter-process communication

This Python application launches a couple of processes which communicate with each other using the publish-subscribe messaging pattern.

There are 5 main objects :

• MotionDetector (Publisher) : Gets a message from the Flask API, creates an object of the class MotionVector and adds this object to the queue
• SingleShotDetector (Publisher-Subcriber) : Gets the messages from the MotionVector's queue, process them and puts them in the DetectionVector's queue.
• Logger (Subscriber) : Reads messages from the both queues and logs them to the standard output.
• MotionVector (Topic) : Class which allows us to create messages under a specific format
• DetectionVector (Topic) : Class which allows us to create messages under a specific format

To be able to correctly share data between processes, the decision to use Queues was made.
Queues are using pipes and a few locks/semaphores, so this guarantees us that data will not be corrupted.

Test the application

Flask is using PORT 5000, make sure that the port is not already in use.

• 1 : Open the folder inter_process_communication in a terminal window

• 2 : Copy and past this shell command in the terminal window : docker-compose -f devops/docker_compose_inter_process_comm.yml up --build
  If everything went good you should see something similar to this in your terminal window :

• 3 : It's time to open the Swagger ! (API's documentation)
  Copy and paste this link into your web browser : http://0.0.0.0:5000/doc The swagger must open in your web browser :

• 4 : Click on "POST" row, it will open, click on try it out
  Write a message and click on Execute If everything went good flask has returned 201 : created

In your terminal window you should see this message printed 2 times, logger consumes 2 queues, first the message will be taken from the first queue and printed and then when the process SingleShotDetector will process the message it will exist in the second queue and the logger will print it a second time.

To stop the docker container :

Please use the command docker stop $(docker ps -a -q) to stop the application.
If you will try to stop it only with the CTRL + C it won't stop correctly as the application has not only one process running.