Container name doesnt resolve when using another container as Network Mode

General
1

Running on Ubuntu 22.04.2
Docker version 25.0.3
Im having issues resolving container name through DNS when using another container as the network mode

Sorry if this is a stupid question, but Ive been hitting a brick wall when researching this topic. I have a media server Ive written up in compose that runs a vpn, bunch of *arrs, qbittorrent, and some media organization. In this particular example I have the vpn, sonarr, qbittorrent, and Jellyfin.

Docker-compose here

vpn-network is external and created as a standard bridge network, I know there are issues resolving container names with the default network.

So the issue Im having is when connecting between apps. To connect Sonarr to Qbittorrent, I use localhost:8085 as they are both using Proton as their network mode and all the ports are exposed, works great.

My issue is when I want to connect from Sonarr to Jellyfin. Sonarr is not able to send a test to jellyfin saying it cant resolve the name

Unable to send test message: Name does not resolve (jellyfin:8096)

However if I get the actual ip when checking the vpn-network and plug that in, its able to find Jellyfin.

Some other information, when running ‘docker exec -it sonarr cat /etc/resolv.conf’ or for vpn I see

nameserver 127.0.0.1
options edns0 trust-ad ndots:0

and when running ‘docker exec -it jellyfin cat /etc/resolv.conf’ I get

nameserver 127.0.0.11
options edns0 trust-ad ndots:0

So what’s the deal with the DNS, I assume this is because I am using network_mode to use the vpn container. Not only that but since it is a vpn with a connection outwards. I know I dont have to use a vpn with sonarr but I would like to keep everything tucked behind a vpn for privacy reasons.

I’ve yet to completely rule out that this is just a sonarr bug, but considering its explicitly saying the name doesn’t resolve I don’t think it is.
I can confirm that DNS works for other apps, I have nginxproxymanager running that routes traffic to Jellyfin and other services I have running and that uses container names completely fine, no problems there.

Obviously I can look up the ip and use that, but thats going to change when I restart things, make changes, etc and I dont want to have to manually update all these things. Using container names is so much more reliable.

2

Of course after dealing with this for a week or so and finally deciding to make some posts, I figured it out last night. This Gluetun image I am using (qmcgaw/gluetun) for my vpn actually has a DNS_ADDRESS env variable that by default is set to 127.0.0.1 . Literally just set that to DNS_ADDRESS=127.0.0.11 (the address for Docker’s DNS) and everything works as expected again. As expected, I was dumb.