docker-urbit

Dockerfile for Urbit. Built image is paulbellamy/urbit.

Usage

To start a new urbit with a pier called mycomet:

$ docker run -ti paulbellamy/urbit -c mycomet

Don't forget -t and -i or urbit will shut down immediately due to the lack of an attached TTY.

HTTP access

The image exposes the urbit's http server at port 8443. Publishing that port Will allow you to see your urbit's hosted content at localhost:8080:

$ docker run -ti -p 8080:8443 paulbellamy/urbit -c mycomet

Port 8443 requires authenticateion, which makes sense since the docker image may be running on a server. To publish the local pre-authenticated port (8080), and only make it accessible locally, you can start your container like this:

$ docker run -ti -p 127.0.0.1:8080:8080 paulbellamy/urbit -c mycomet

Binding data to the host FS

The container's working directory is /urbit, and all data is stored there, so if you want to keep your data, you'll want to mount your urbit from the local filesystem. For example, to create a comet in a directory mycomet:

$ docker run --rm \
    --volume `pwd`:/urbit \
    -ti paulbellamy/urbit \
    -c mycomet

You will most certainly want to access this data with a non-root account, but urbit is running as root inside this container (for a few good reasons). To ensure that you can access the files and directories created by urbit, run the container with the --user option set to your user and group id:

$ docker run --rm \
    --volume `pwd`:/urbit \
    --user `id -u`:`id -g` \
    -ti paulbellamy/urbit \
    -c mycomet

Currently you need to mount the whole /urbit when creating a new urbit, because if you mount to /urbit/mycomet, docker will initialize that directory and vere wont start if it exists (even though it will be empty).

If you're booting an existing urbit however, you'll want to mount only the pier into /urbit/<name>. For example, to launch an existing urbit fintud-macrep:

$ docker run \
    --volume `pwd`/fintud-macrep:/urbit/fintud-macrep \
    -ti paulbellamy/urbit \
    -c mycomet

Note the --rm in the -c examples above, most likely you will want to run a container with -c once, then start another, permanent one to keep it running.

Starting with a fake ~zod

This can be useful for development. urbit/arvo needs to be cloned to bootstrap the new zod.

$ git clone https://github.com/urbit/arvo.git
$ docker run -ti -v `pwd`/arvo:/arvo paulbellamy/urbit -F -I zod -A /arvo -c myzod

There is a tag fakezod with a different entrypoint and the arvo repo checked out in /arvo to make this even easier:

$ docker run -ti paulbellamy/urbit:fakezod -A /arvo -c myzod

to run a container from an existing fakezod pier, omit -A and -c:

$ docker run -ti -v `pwd`/myzod:/urbit/myzod paulbellamy/urbit:fakezod myzod

Troubleshooting

If your container keeps printing something like

unix: stopping process 1, live in fzod...

and restarting, you are stuck in a restart loop. Your urbit thinks another process is running, and by trying to kill it terminates itself.

The fix is rather simple. First, make sure your container is stopped:

$ docker kill fakezod

then simply delete the vere lockfile:

$ rm /path/to/pier/.vere.lock

if you dont have the volume mounted, docker exec in the running container:

$ docker exec fakezod rm /urbit/fzod/.vere.lock

More Info

For more info on usage, please see the urbit setup docs.