Improve Configuration Management

[JochenKienzle]

At the moment, the configuration for the docker images (e.g. via docker-compose) is split between different repositories and not all configuration parameters can be set/overwritten (or not clear how to do it). Also some hidden behavior could happen: e.g. the .dockerignore in the iot-event-analytics submodule could avoid you to include a config folder in the image itself.

The proposal:

• All dynamic parts (folder-paths, network-names, port-bindings, tags, specific component-configs, ...) for each container is configure-able (in an easy way)
• Main configuration (e.g. *.env') to set/overwrite these parameters.
• Common / general configuration: e.g. network-name, protocol-gateway, ...
• Different 'deployment' modes available: E.g. all in one image (use COPY) or to bind a volume from the host filesystem.

[llakie]

.dockerignore should facilitate the process of using the COPY command in the Dockerfiles. Everything should be ignored, which is not needed. What do you mean by "the configuration is split between different repositories"? The whole configuration of the stack is provided in the vehicle-edge project under docker-compose/config or setup/general.

[llakie]

We should distinguish between container runtime properties and application properties The first can be done by making the yml use more environment variables and the latter is part of the application implementation.