[Core] Add internal port feature

[Bockiii]

This fixes #2270

The addition of the feature does not change any behavior for current users. You have to manually add the environment variable "PORT" to the container to use this internal feature.

What it does:
If you add "PORT" to your docker container environment call, the script replaces the default port 80 in the config files to whichever port you assign.

I will add the how-to to the wiki as soon as it's merged.

[em92]

Only 80 port is exposed. If you switch it to other one, you won't be able to connect to it.

[em92]

Closing, as parent issue is resolvable without additional changes.

[Bockiii]

No, you misunderstood the query.

If you are using other network modes than "bridged" in docker, the environmental mapping is not available for some. So for example, if you want 2 containers in their own DMZ, there is no mapping at all between those containers and the host, but all ports are open between them. As if they were one machine.

The problem comes in when you have two containers that both use the same port (for example rss-bridge and a webserver container). Both listen on 80. If you now try to connect from a third container inside this DMZ to port 80, it doesn't know where to connect to.

It's a niche case, but still valid. One would be the case that the Issue opener provided: VPN container with nginx and rss-bridge.

In these networking scenarios, "exposed" also doesn't do anything. All ports are open inside the dmz.

[Bockiii]

On a sidenote: The "expose" feature is helpful for automated environments. You can say "auto-map exposed ports" and some container runtimes automatically attach a random port to the exposed ports, but in the end its just that: A help for automation. You can still map whatever port your heart desires to the container, doesn't matter if it's exposed or not. So changing the internal port to 5555, you only need to also map that port to the outside. eg docker run -e PORT=5555 -p 5555:5555 rss-bridge

[em92]

Sorry for misunderstanding the case and thank you for explanation!

I changed environment variable from PORT to HTTP_PORT as it is more commonly used. For example here https://symfony.com/doc/4.4/configuration/env_var_processors.html

[Bockiii]

If we want to switch the name, i would go for a different wording. I had the same issue with another application:

GoogleChrome/lighthouse-ci#258 (comment)

Rancher (think of it as kubernetes) auto creates a variable called SERVICE_PORT. So if someone uses Rancher and also has a service called "HTTP", theres going to be problems.

We can keep it like this for now since you already merged it. Just be aware of this if someone else opens an issue :)

[Bockiii]

tbh, its probably unlikely that someone names their kubernetes service "HTTP". Just be aware :D