Switch to using one Compose recipe with multiple profiles/flavours

[alyssadai]

By having tools which are supposed to work together in different Docker networks, we currently often need to jump through hoops to make the networking work correctly so that the containers can communicate with one another using their public IPs/URLs.

Meanwhile, the point of having cooperative containers all in the same network is to avoid the hassle of having to do this, as containers can simply refer to one another by their service name. Having all tools be part of the same Docker network by default will make setup a lot less brittle compared to having to potentially manage multiple networks at once.

We should switch to a single docker-compose.yml with different profiles (see https://stackoverflow.com/a/65957695 for example) replacing our separate .yml files for each deployment flavour we support.

context: https://docs.docker.com/compose/profiles/

• update README.md with relevant info about this recipe e.g., run commands with profiles

To close ensure that:

• admin password et al are ENV variables
• profiles are set up so we can run the different flavours (everything currently done via discrete docker compose file) - or any required combo
• check if there is a clever way to have a "default" profile that is launched if no profile is explicitly provide
• write output of setup.sh to default named log file

To check:

• do we want to keep .env and local_nb_nodes.json as templates? If so, should we make the repo gitignore them?
• setting up two nodes to test the f-API? won't be addressed in this issue
• add note (in docs, maybe) that data path needs to be changed

[neurobagel-bot]

🚀 Issue was released in v0.0.1 🚀