[Python] Is there a low-maintenance way to implement microservice chassis pattern in Python?

Context: I'm developing my first microservice project, and still trying to wrap my head around some best practices.

The project structure is not that complicated:

DaProject/
+- ServiceA/
| +- main.py
| \- Dockerfile
+- ServiceB/
| +- some_ServiceB_specific_lib/
| +- main.py
| \- Dockerfile
\- docker-compose.yml


Now, as I played around with implementing a couple of services, I've figured out some of the cross-cutting concerns for my project. Since all services will be implemented in Python, I want to take care of those concerns using Service Chassis pattern. The question is, is there a simple, but decent way to do that? For example, if I do this:

DaProject/
+- Chassis/
| \- __init__.py
+- ServiceA/
| +- main.py
| \- Dockerfile
+- ServiceB/
| +- some_ServiceB_specific_lib/
| +- main.py
| \- Dockerfile
\- docker-compose.yml


then I will need to reference the Chassis package from within the services.

What I have thought about:

1. I can't just reference ../Chassis in any Dockerfile, as it is outside of the build scope. Okay, that makes sense.
2. I don't want to manually copy the chassis into each service directory whenever I happen to update it. It's a very simple solution, but high-maintenance and error-prone.
3. Hardlinking the directory is a hack, and will likely break if the code is moved somewhere else.
4. Implementing a local pypi repo sounds like a high-maintenance way to handle it. Perhaps it would be more worth it I was implementing a more abstract and reusable library, not something as project-bound as this.
5. Googling pointed me towards 'git submodule' concept... but I have only basic familiarity with Git, and so I can't really judge how complex this solution is.

Is there any other way I'm not aware of? Or should I bite the bullet and learn the submodule feature?

Continue reading...