Weird issue with two containers with the same internal port

Traefik Traefik v2
1

Hi, just finished setting up traefik, ran into the usual issues that pop up when installing a new service ^^ was able to solve everything with the documentation though. Except for one weird issue:

I'm running (amongst others) two containers in docker, jellyfin and emby, which both have the same internal port, 8096. Jellyfin is mapped 8096:8096, emby is mapped 8092:8096.

Now, traefik doesn't recognize emby correctly through the automatic detection. It tries to connect to the service through [IP-docker-network]:1900 for some reason I don't understand.

When I manually define a router and service for emby in the fileConfig.yml for traefik, I can connect to it. It took me some time to get there though, as I didn't realize I had to map the internal port for the service, not the external one I expected.

As I'd really prefer to use the automatic discovery though, I'm wondering: Is there an issue due to the fact that both containers use the same internal port? Does traefik allow for such use cases?

If that helps, I'm running both containers on their own docker network, both are also connected to traefik_proxy_network which traefik runs in.

2

When using Traefik you don't need any port mapping with Configuration Discovery, the internal requests should go through a Docker network. Otherwise you can circumvent all security measures you set up with Traefik middlewares.

Traefik tries to auto-detect the target port by checking expose from Dockerfile, you can overwrite it with the correct port, see simple Traefik example.

  whoami:
    image: traefik/whoami:v1.10
    networks:
      - proxy
    labels:
      - traefik.enable=true
      - traefik.http.routers.mywhoami.rule=Host(`whoami.example.com`) || Host(`www.whoami.example.com`)
      - traefik.http.services.mywhoami.loadbalancer.server.port=80