DNM/WIP dont try to listen on the advertise address

[dcd]

• sometimes docker might not be in use, but you still want to tell clients a different IP than you bind to inside the container

@tgorgdotcom this is here for your consideration, i am not really a python guy and haven't put a ton of thought into this yet. however, the current is_docker function is also breaking for k8s (even when actually using docker), and there is no guarantee that docker will be the CRI.

[dcd]

so the first push didnt actually work , i made a cheap workaround to get service back, but i doubt its how you want it in the end.

the is_docker check does not work on k8s, and docker may not be the container runtime in use anyways.

i think the preferred solution will probably be to introduce a new parameter for listen_address, but i also dont understand the use case where a user would not desire 0.0.0.0

[deathbybandaid]

#92 has some stuff built in to address this

[tgorgdotcom]

Hello dcd. I think I'll incorporate this manually as I'm working on some other issues. One thought I wanted to run by you -- What would you think of the following logic:

• add bind_ip and bind_port to main config
• change is_docker to rely on a file created by the Dockerfile (RUN touch is_docker or the like)
• when we need to figure out what port to bind to, rely on the following rules:
  • If bind_ip and bind_port are set, use these settings
  • If bind_ip/bind_port is not set use our default way to figure out ports:
    • If using docker, rely on 0.0.0.0
    • If not using docker, rely on the plex_accessible_ip/plex_accessible_port

I want to change plex_accessible_ip/plex_accessible_port to advertise_ip/advertise_post like you did, but I want to make sure we don't do any settings-breaking changes until a major release. Let me know if this is good logic for your setup.

[deathbybandaid]

For fHDHR, I use address and discovery_address.
SSDP does no good with 0.0.0.0

It will try to use discovery_address, and if not set, it will try to use address. If whichever value it uses is 0.0.0.0 ssdp simply does not run.

If the config defaults to 0.0.0.0 for address, and docker users don't override that in their config file, then the is_docker() is not really needed.

#101 goes hand in hand with this. If 0.0.0.0, the lineup won't work well. Gathering the base_url at the time of client connect is an easy way for 0.0.0.0 to work properly.

Honestly, for fHDHR, runnng locally, I purposely don't use ssdp. It only 'helps' with auto discovery in Plex. It's easy enough to manually insert the ip:port and not run an extra thread.

[tgorgdotcom]

Right. bind_ip/bind_port would only be used as the listen address, not the address returned in an SSDP reply message or the value in . I think you're right about the is_docker function -- we probably don't need it.

So maybe it should be more like this:

• add bind_ip and bind_port to main config
• remove is_docker function entirely
• when we need to figure out what IP/port to listen on, rely on the following rules:
  • If bind_ip and bind_port are set, use these settings
  • If not, default to 0.0.0.0/6077
• when we need to figure out what IP/port to report as the server IP, rely on the following rules:
  • use advertise_ip (renamed from plex_accessible_ip) for ip
  • use advertise_post (renamed from plex_accessible_port) or 6077 for port

I'll try to see if I can change it in 2.x, as this changes some config names and meanings

[deathbybandaid]

I don't think you need 2 port settings