norns-test-dummy

This image runs the full Norns stack (crone, maiden and matron), but without any sound output. It uses the dummy jackd sound driver.

Please view the Dockerfile to see the steps taken. In particular note that we've created a we user account and home directory.

It is possible to use this image to get sound output and grid functionality on Linux.

Quick start

Run the following (it is not necessary to have cloned this repo):

docker run --rm -it \
    --ulimit rtprio=95 --ulimit memlock=-1 --shm-size=256m \
    -p 5000:5000 -p 5555:5555 -p 5556:5556 \
    samdoshi/norns-test-dummy

Then visit http://127.0.0.1:5000/maiden/ in your browser.

The above command will pull a copy of samdoshi/norns-test-dummy from Docker Hub if it doesn't exist locally. See below for details to build it yourself.

Type Ctrl-b d to quit tmux and the Docker session.

Summary of Makefile targets

Please view the Makefile to see the exact options given to the docker commands.

• build: builds the Docker image and tags it as samdoshi/norns-test-dummy.
• run: runs the image leaving you in a tmux session with jackd, crone, matron and maiden running in individual panes.
• run-bash: runs the image but doesn't start the tmux session.
• run-bash-root: as run-bash but starts a root bash shell, instead of the we user.
• shell: starts a shell in an already running Docker process, it is useful with the run target.
• shell-root: as shell but starts as root.

Getting sound output on Linux with ALSA pass-through

You can also use this image to produce audio on Linux, as it's running in a container and not a virtual machine, performance should be nearly identical to running it on the host OS. However we will need to dedicate a sound card to the image (i.e. it cannot be in use by either Jack or PulseAudio).

Broadly speaking we need to do the following:

1. Share /dev/snd with the container.

   This is achieved by adding --device /dev/snd to the docker run command.

2. Provide a new /etc/jackdrc to the container.

   Either we can create a new Dockerfile deriving from this, or we can use the -v option with docker run to provide the new file from the local filesystem.

3. Ensure that the we user has the correct permissions to access the sound card.

   This is achieved by adding --group-add <gid of host audio group> to the docker run command.

As an example with my own setup:

/tmp/jackdrc:

/usr/local/bin/jackd -R -d alsa -d hw:Crimson

(I recommend using hardware names, you can find them by running cat /proc/asound/cards)

Then I can use the following docker run command:

docker run --rm -it \
    --device /dev/snd \
    --group-add $(getent group audio | cut -d: -f3) \
    --ulimit rtprio=95 --ulimit memlock=-1 --shm-size=256m \
    -p 5000:5000 \
    -p 5555:5555 \
    -p 5556:5556 \
    -v /tmp/jackdrc:/etc/jackdrc \
    --name norns \
    samdoshi/norns-test-dummy

(We use the $(getend...) as we need to use the gid of the group from the host)

If all has gone well, you'll see the 4 panes inside tmux appear, and once all the applications have started running you can visit http://127.0.0.1:5000/maiden/ and run a script. I would suggest starting with tehn/awake.lua.

Connecting a grid with Linux

This is done in a similar way to getting ALSA sound working.

1. Share the grid serial device with the container.

2. Ensure that the we user has permissions to access the device.

We need to determine what the grid device is, and also what group it's owned by.

Firstly, the easiest way to determine the device path is running dmesg just after plugging it in:

$ dmesg
...
[22512.286363] usb 1-3.3: Detected FT232RL
[22512.286593] usb 1-3.3: FTDI USB Serial Device converter now attached to ttyUSB0

Here we can see that the device is attached to ttyUSB0, which is actually at /dev/ttyUSB0.

To determine the group:

$ ls -la /dev/ttyUSB0
crw-rw---- 1 root uucp 188, 0 Sep  8 14:08 /dev/ttyUSB0

We can see that the group is uucp, this is distro specific, the example given is from my computer running Arch Linux.

Using that information we can add the following to a docker run command to pass the grid through: --device /dev/ttyUSB0 --group-add $(getent group uucp | cut -d: -f3)

For my own personal setup (and combined with the audio from above), my docker run command becomes:

docker run --rm -it \
    --device /dev/snd \
    --group-add $(getent group audio | cut -d: -f3) \
    --device /dev/ttyUSB0 \
    --group-add $(getent group uucp | cut -d: -f3) \
    --ulimit rtprio=95 --ulimit memlock=-1 --shm-size=256m \
    -p 5000:5000 \
    -p 5555:5555 \
    -p 5556:5556 \
    -v /tmp/jackdrc:/etc/jackdrc \
    --name norns \
    samdoshi/norns-test-dummy

I recommend running the tehn/earthsea.lua script to test the grid functionality.