Skip to content

Instantly share code, notes, and snippets.

@sunsided
Created February 29, 2020 13:48
Show Gist options
  • Save sunsided/7840e89ff4e11b64a2d7503fafa0290c to your computer and use it in GitHub Desktop.
Save sunsided/7840e89ff4e11b64a2d7503fafa0290c to your computer and use it in GitHub Desktop.
WIFIonICE vs Docker: Fixing DB (Deutsche Bahn) WIFI by moving Docker away from 172.18.x.x in /etc/docker/daemon.json
{
"bip": "172.39.1.5/24",
"fixed-cidr": "172.39.1.0/25",
"runtimes": {
"nvidia": {
"path": "nvidia-container-runtime",
"runtimeArgs": []
}
}
}
@sunsided
Copy link
Author

sunsided commented Oct 1, 2021

@arne-cl Just to clarify: changing the bip and fixed-cidr subnet ranges didn't work? For the fix, the file is assumed to stay identical otherwise.

@arne-cl
Copy link

arne-cl commented Oct 4, 2021

Dear @sunsided,
I deleted existing networks with docker network rm ... first, then saved your snippet verbatim in /etc/docker/daemon.json (I didn't have that file before).

Afterwards, docker wouldn't work, i.e.

docker run hello-world
docker: Cannot connect to the Docker daemon at unix:///var/run/docker.sock. Is the docker daemon running?.
See 'docker run --help'.

Running sudo systemctl restart docker.service didn't help either. I also tried it without the nvidia part:

{
    "bip": "172.39.1.5/24",
    "fixed-cidr": "172.39.1.0/25"
}

but the result was the same.

@henningsway
Copy link

it just worked for me, thanks!

@malte-j
Copy link

malte-j commented May 6, 2022

Removing all networks first and leaving out the Nvidia config worked for me, thanks!

@orzechow
Copy link

orzechow commented Apr 4, 2023

With the above solution, each time when connecting to WIFIonICE I was forced to remove networks created by docker compose again – which invalidated images and lead to running docker prune way too often…

This works for a couple of months now, without the need to prune networks each time (only once after initial setup):

/etc/docker/daemon.json

{
    "bip": "172.39.1.5/24",
    "fixed-cidr": "172.39.1.0/25",
    "default-address-pools":
    [{
        "base": "172.39.0.0/16",
        "size": 24
    }],
    "runtimes": {
        "nvidia": {
            "path": "nvidia-container-runtime",
            "runtimeArgs": []
        }
    }
}

@NiklasMM
Copy link

NiklasMM commented Apr 28, 2023

Doesn't work for me =(

The address for docker0 complies, but the br-<somethin> networks continue to use the 172.18.x.x range

Docker version 23.0.4, build f480fb1

Edit:

Ok, just got it working by running docker network prune. Networks created after that will use the new range

@orzechow
Copy link

Nice! 🚀

@leonp5
Copy link

leonp5 commented Jul 17, 2023

I had to delete all already existing container and images. After i did that, it worked.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment