Shortly after writing this, I realized that ZwaveJS2Mqtt can be used with a websocket connection instead of the mqtt. I've had a great experience with it, so there's really no good reason to use what is below.
Documentation: https://zwave-js.github.io/zwavejs2mqtt/#/
I'll leave the text below, because somebody might find it useful.
Until a better solution comes around, this is how I got zwave-js running in docker for use in home assistant. The timing was just right for me to create a new network, as my first z-wave devices only arrive a few days before the Home Assitant annoucement.
I choose to run my containers with named local volumes, host networking, and automatic watchtower updates. Therefore, take this into account when following these instructions.
Launch python
import random
bit_choices = ["0x{:02d}".format(x) for x in range(0,16)]
key = ", ".join([random.choice(bit_choices) for x in range(0,16)])
print(key)
save this key to a safe location. Set it as a variable for now
NETWORK_KEY="<script_output_here>"
mkdir $HOME/zwavejs-image
cd $HOME/zwavejs-image
git clone https://github.com/kpine/dockerfiles.git
cd dockerfiles/zwave-js-server
docker build -t zwavejs .
docker volume create zwavejs-cache
docker volume create zwavejs-logs
Note that network mode is set to host. You may want to add this container to a network or publish a port if this doesn't work for you. by default, zwave-js runs on port
3000
Note that I'm mounting my zwave stick to /dev/zwave
. This is the default path set in zwave-js code, but it can be configured otherwise.
docker run --name=zwavejs \
--device=/dev/serial/by-id/usb-Silicon_Labs_HubZ_Smart_Home_Controller_81300326-if00-port0:/dev/zwave \
--network=host \
-e NETWORK_KEY=${NETWORK_KEY} \
-l com.centurylinklabs.watchtower.enable=false \
-v zwavejs-cache:/cache \
-v zwavejs-logs:/logs \
--restart=unless-stopped -d \
zwavejs
If you're doing this in docker compose:
great guide!