This project describes some of the many ways Node-RED can be run under Docker and has support for multiple architectures (amd64, arm32v6, arm32v7 and arm64v8). Some basic familiarity with Docker and the Docker Command Line is assumed.
This project also provides the build for the nodered/node-red-docker containers on DockerHub.
To run this directly in Docker at it's simplest just run:
docker run -it -p 1880:1880 --name mynodered nodered/node-red-docker
Let's dissect that command:
docker run - run this container... and build locally if necessary first.
-it - attach a terminal session so we can see what is going on
-p 1880:1880 - connect local port 1880 to the exposed internal port 1880
--name mynodered - give this machine a friendly local name
nodered/node-red-docker - the image to base it on - currently Node-RED v0.20.7
Running that command should give a terminal window with a running instance of Node-RED.
Welcome to Node-RED
===================
10 Aug 12:57:10 - [info] Node-RED version: v0.20.7
10 Aug 12:57:10 - [info] Node.js version: v10.16.2
10 Aug 12:57:10 - [info] Linux 4.19.58-v7+ arm LE
10 Aug 12:57:11 - [info] Loading palette nodes
10 Aug 12:57:16 - [info] Settings file : /data/settings.js
10 Aug 12:57:16 - [info] Context store : 'default' [module=memory]
10 Aug 12:57:16 - [info] User directory : /data
10 Aug 12:57:16 - [warn] Projects disabled : editorTheme.projects.enabled=false
10 Aug 12:57:16 - [info] Flows file : /data/flows.json
10 Aug 12:57:16 - [info] Creating new flow file
10 Aug 12:57:17 - [warn]
---------------------------------------------------------------------
Your flow credentials file is encrypted using a system-generated key.
If the system-generated key is lost for any reason, your credentials
file will not be recoverable, you will have to delete it and re-enter
your credentials.
You should set your own key using the 'credentialSecret' option in
your settings file. Node-RED will then re-encrypt your credentials
file using your chosen key the next time you deploy a change.
---------------------------------------------------------------------
10 Aug 12:57:17 - [info] Server now running at http://127.0.0.1:1880/
[...]
You can then browse to http://{host-ip}:1880 to get the familiar Node-RED desktop.
The advantage of doing this is that by giving it a name we can manipulate it more easily, and by fixing the host port we know we are on familiar ground. (Of course this does mean we can only run one instance at a time... but one step at a time folks...)
If we are happy with what we see we can detach the terminal with Ctrl-p``Ctrl-q - the container will keep running in the background.
To reattach to the terminal (to see logging) run:
$ docker attach mynodered
If you need to restart the container (e.g. after a reboot or restart of the Docker daemon):
$ docker start mynodered
and stop it again when required:
$ docker stop mynodered
Note: this Dockerfile is configured to store the flows.json file and any extra nodes you install "outside" of the container. We do this so that you may rebuild the underlying container without permanently losing all of your customisations.
The Node-RED images come in different variations and are supported by manifest lists (auto-detect architecture). This makes it more easy to deploy in a multi architecture Docker environment. E.g. a Docker Swarm with mix of Raspberry Pi's and amd64 nodes.
The tag naming convention is <node-red-version>-<node-version>-<image-type>-<architecture>, where:
<node-red-version>is the Node-RED version.<node-version>is the Node JS version.<image-type>is type of image and is optional, can be either none, minimal devtools or python2.- none : is the default and has Python 3 installed
- minimal : has no Python installed
- devtools : has Python 3 installed and has development tools installed. currently build-base only
- python2 : has Python 2 installed
<architecture>is the architecture of the Docker host system, can be either amd64, arm32v6, arm32v7, arm64.
The Node-RED images are based on official Node JS Alpine Linux images to keep them as small as possible. Using Alpine Linux reduces the built image size, but removes standard dependencies that are required for native module compilation. If you want to add dependencies with native dependencies, extend the Node-RED image with the missing packages on running containers or build new images see docker-custom.
The following table shows the variation of provided Node-RED images.
| Tag | Node | Arch | Python | GPIO | Dev | Base Image |
|---|---|---|---|---|---|---|
| 0.20.7-10-amd64 | 10 | amd64 | 3.x | no | no | amd64/node:10-alpine |
| 0.20.7-10-arm32v6 | 10 | arm32v6 | 3.x | yes | no | arm32v6/node:10-alpine |
| 0.20.7-10-arm32v7 | 10 | arm32v7 | 3.x | yes | no | arm32v7/node:10-alpine |
| 0.20.7-10-arm64v8 | 10 | arm64v8 | 3.x | no | no | arm64v8/node:10-alpine |
| 0.20.7-10-minimal-amd64 | 10 | amd64 | no | no | no | amd64/node:10-alpine |
| 0.20.7-10-minimal-arm32v6 | 10 | arm32v6 | no | no | no | arm32v6/node:10-alpine |
| 0.20.7-10-minimal-arm32v7 | 10 | arm32v7 | no | no | no | arm32v7/node:10-alpine |
| 0.20.7-10-minimal-arm64v8 | 10 | arm64v8 | no | no | no | arm64v8/node:10-alpine |
| 0.20.7-10-devtools-amd64 | 10 | amd64 | 3.x | no | yes | amd64/node:10-alpine |
| 0.20.7-10-devtools-arm32v6 | 10 | arm32v6 | 3.x | yes | yes | arm32v6/node:10-alpine |
| 0.20.7-10-devtools-arm32v7 | 10 | arm32v7 | 3.x | yes | yes | arm32v7/node:10-alpine |
| 0.20.7-10-devtools-arm64v8 | 10 | arm64v8 | 3.x | no | yes | arm64v8/node:10-alpine |
| 0.20.7-10-python2-amd64 | 10 | amd64 | 2.x | no | no | amd64/node:10-alpine |
| 0.20.7-10-python2-arm32v6 | 10 | arm32v6 | 2.x | yes | no | arm32v6/node:10-alpine |
| 0.20.7-10-python2-arm32v7 | 10 | arm32v7 | 2.x | yes | no | arm32v7/node:10-alpine |
| 0.20.7-10-python2-arm64v8 | 10 | arm64v8 | 2.x | no | no | arm64v8/node:10-alpine |
| Tag | Node | Arch | Python | GPIO | Dev | Base Image |
|---|---|---|---|---|---|---|
| 0.20.7-12-amd64 | 12 | amd64 | 3.x | no | no | amd64/node:12-alpine |
| 0.20.7-12-arm32v6 | 12 | arm32v6 | 3.x | yes | no | arm32v6/node:12-alpine |
| 0.20.7-12-arm32v7 | 12 | arm32v7 | 3.x | yes | no | arm32v7/node:12-alpine |
| 0.20.7-12-arm64v8 | 12 | arm64v8 | 3.x | no | no | arm64v8/node:12-alpine |
| 0.20.7-12-minimal-amd64 | 12 | amd64 | no | no | no | amd64/node:12-alpine |
| 0.20.7-12-minimal-arm32v6 | 12 | arm32v6 | no | no | no | arm32v6/node:12-alpine |
| 0.20.7-12-minimal-arm32v7 | 12 | arm32v7 | no | no | no | arm32v7/node:12-alpine |
| 0.20.7-12-minimal-arm64v8 | 12 | arm64v8 | no | no | no | arm64v8/node:12-alpine |
| 0.20.7-12-devtools-amd64 | 12 | amd64 | 3.x | no | yes | amd64/node:12-alpine |
| 0.20.7-12-devtools-arm32v6 | 12 | arm32v6 | 3.x | yes | yes | arm32v6/node:12-alpine |
| 0.20.7-12-devtools-arm32v7 | 12 | arm32v7 | 3.x | yes | yes | arm32v7/node:12-alpine |
| 0.20.7-12-devtools-arm64v8 | 12 | arm64v8 | 3.x | no | yes | arm64v8/node:12-alpine |
| 0.20.7-12-python2-amd64 | 12 | amd64 | 2.x | no | no | amd64/node:12-alpine |
| 0.20.7-12-python2-arm32v6 | 12 | arm32v6 | 2.x | yes | no | arm32v6/node:12-alpine |
| 0.20.7-12-python2-arm32v7 | 12 | arm32v7 | 2.x | yes | no | arm32v7/node:12-alpine |
| 0.20.7-12-python2-arm64v8 | 12 | arm64v8 | 2.x | no | no | arm64v8/node:12-alpine |
The Node-RED images have either no Python, Python 3.x or Python 2.x pre-installed and for arm32v6 and arm32v7 Node-RED build-in GPIO enabled.
- All images have bash, tzdata, nano, curl git and openssl tools pre-installed to support Node-REDs Projects feature.
- note: Python 2.7 reaches end of life on January 1st, 2020! Therefore it's highly recommended to use Python 3 based images, if you need Python pre-installed.
The following table shows the provided Manifest Lists.
| Tag | Node-RED Base Image |
|---|---|
| latest, 0.20.7, | raymondmm/0.20.7-10-amd64 |
| latest-10, 0.20.7-10 | raymondmm/0.20.7-10-arm32v6 |
| raymondmm/0.20.7-10-arm32v7 | |
| raymondmm/0.20.7-10-arm64v8 | |
| latest-minimal, 0.20.7-minimal, | raymondmm/0.20.7-10-amd64-minimal |
| latest-10-minimal, 0.20.7-10-minimal | raymondmm/0.20.7-10-arm32v6-minimal |
| raymondmm/0.20.7-10-arm32v7-minimal | |
| raymondmm/0.20.7-10-arm64v8-minimal | |
| latest-devtools, 0.20.7-devtools | raymondmm/0.20.7-10-amd64-devtools |
| latest-10-devtools, 0.20.7-10-devtools | raymondmm/0.20.7-10-arm32v6-devtools |
| raymondmm/0.20.7-10-arm32v7-devtools | |
| raymondmm/0.20.7-10-arm64v8-devtools | |
| latest-python2, 0.20.7-python2 | raymondmm/0.20.7-10-amd64-python2 |
| latest-10-python2, 0.20.7-10-python2 | raymondmm/0.20.7-10-arm32v6-python2 |
| raymondmm/0.20.7-10-arm32v7-python2 | |
| raymondmm/0.20.7-10-arm64v8-python2 |
| Tag | Node-RED Base Image |
|---|---|
| latest-12, 0.20.7-12 | raymondmm/0.20.7-12-amd64 |
| raymondmm/0.20.7-12-arm32v6 | |
| raymondmm/0.20.7-12-arm32v7 | |
| raymondmm/0.20.7-12-arm64v8 | |
| latest-12-minimal, 0.20.7-12-minimal | raymondmm/0.20.7-12-amd64-minimal |
| raymondmm/0.20.7-12-arm32v6-minimal | |
| raymondmm/0.20.7-12-arm32v7-minimal | |
| raymondmm/0.20.7-12-arm64v8-minimal | |
| latest-12-devtools, 0.20.7-12-devtools | raymondmm/0.20.7-12-amd64-devtools |
| raymondmm/0.20.7-12-arm32v6-devtools | |
| raymondmm/0.20.7-12-arm32v7-devtools | |
| raymondmm/0.20.7-12-arm64v8-devtools | |
| latest-12-python2, 0.20.7-12-python2 | raymondmm/0.20.7-12-amd64-python2 |
| raymondmm/0.20.7-12-arm32v6-python2 | |
| raymondmm/0.20.7-12-arm32v7-python2 | |
| raymondmm/0.20.7-12-arm64v8-python2 |
With the support of Docker manifest list, there is no need to explicit add the tag for the architecture to use. When a docker run command or docker service command or docker stack command is executed, docker checks which architecture is required and verifies if it is available in the docker repository. When it does, docker pulls the matching image for it.
Therefor all tags regarding Raspberry PI's are dropped.
For example: suppose you are running on a Raspberry PI 3B, which has arm32v7 as architecture. Then just simple run the following command to pull the image (tagged by 0.20.7-10-arm32v7) and run the container.
docker run -it -p 1880:1880 --name mynodered nodered/node-red-docker:latest
The same command can be used for running on an amd64 system, since docker discovers its running on a amd64 host and pulls the image with matching tag (0.20.7-10-amd64).
This gives the advantaged that you don't need to know/specifiy which architecture you are running on and makes docker run commands and docker compose files for flexible and exchangeable across systems.
The non-minimal Node-RED images arm32v6 and arm32v7 provide build-in GPIO support, however it is highly recommended to use node-red-node-pi-gpiod instead.
Disadvantages of the build-in GPIO support are:
- Your Docker container needs to be deployed on the same Docker node/host on which you want to control the gpio's.
- Gain access to /dev/mem of your Docker node/host
- privileged=true is not supported for
docker stackcommand
node-red-node-pi-gpiod solves all these disadvantages. With node-red-node-pi-gpiod it is even possible to interact with gpio's of multiple Raspberry Pi's from a single Node-RED container.
If you still do want to make use of the Node-RED build-in GPIO support, run your container like this:
docker run -it --rm -p1880:1880 --user=root --privileged=true -v /dev/mem:/dev/mem nodered/node-red-docker:latest
To save your Node-RED user directory inside the container to a host directory outside the container, you can use the command below. But to allow access to this host directory, the node-red user (default uid=1001) inside the container must have the same uid as the owner of the host directory. To override the default uid and gid of the node-red user inside the the container you can use the option --user="<my_host_uid>:<my_host_gid>":
$ docker run -it --user="<my_host_uid>:<my_host_gid>" -p 1880:1880 -v <host_directory>:/data --name mynodered nodered/node-red-docker
Use case ...
- Suppose you are running on a Raspberry PI with a user named 'pi' and group 'pi'.
$ whoami
- With this user create a directory '~/.node-red'.
$ mkdir ~/.node-red
- Verify newly created directory with:
$ ls -al ~/.node-red
This shows that user pi is owner of this directory:
ls -al ~/.node-red
total 8
drwxr-xr-x 2 pi pi 4096 May 7 20:55 .
drwxr-xr-x 8 pi pi 4096 May 7 20:42 ..
- Now we want to have access to this '~/.node-red' directory with the container so that Node-RED can save user data to it. As we know we need to override the default uid (1001) of the node-red user inside the container with the uid of the pi user. For that we need to know the uid of user pi:
$ id pi
- The uid and gid of user pi are:
uid=1000(pi) gid=1000(pi) [...]
- So the final command becomes:
$ docker run -it --user="1000:1000" -p 1880:1880 -v ~/.node-red:/data --name mynodered nodered/node-red-docker
Running a Node-RED container with a host directory mounted as the data volume,
you can manually run npm install within your host directory. Files created in
the host directory will automatically appear in the container's file system.
Adding extra nodes to the container can be accomplished by running npm install locally.
$ cd ~/.node-red
$ npm install node-red-node-smooth
[email protected] node_modules/node-red-node-smooth
$ docker stop mynodered
$ docker start mynodered
Note : Modules with a native dependencies will be compiled on the host machine's architecture. These modules will not work inside the Node-RED container unless the architecture matches the container's base image. For native modules, it is recommended to install using a local shell or update the project's package.json and re-build.
Below an example of a Docker Compose file which can be run by docker stack or docker-compose.
Please refer to the official Docker pages for more info about docker stack and docker compose.
################################################################################
# Node-Red Stack
################################################################################
#$ docker stack deploy node-red --compose-file docker-compose-node-red.yml
################################################################################
version: 3.7
services:
node-red:
image: nodered/node-red-docker:latest
environment:
- TZ=Europe/Amsterdam
ports:
- "1880:1880"
networks:
- node-red-net
volumes:
- /mnt/docker-cluster/node-red/data:/data
networks:
node-red-net:
The above compose file:
- creates a node-red service
- pulls the latest node-red image
- sets the timezone to Europe/Amsterdam
- Maps the container port 1880 to the the host port 1880
- creates a node-red-net network and attaches the container to this network
- persists the
/datadir inside the container to the/mnt/docker-cluster/node-red/datadir outside the container
This repository contains Dockerfiles to build the Node-RED Docker images listed above.
The package.json is a metafile that downloads and installs the required version
of Node-RED and any other npms you wish to install at build time. During the
Docker build process, the dependencies are installed under /usr/src/node-red.
The main sections to modify are
"dependencies": {
"node-red": "0.20.x", <-- set the version of Node-RED here
"node-red-node-rbe": "*" <-- add any extra npm packages here
},
This is where you can pre-define any extra nodes you want installed every time by default, and then
"scripts" : {
"start": "node-red -v $FLOWS"
},
This is the command that starts Node-RED when the container is run.
Node-RED is started using NPM start from this /usr/src/node-red, with the --userDir
parameter pointing to the /data directory on the container.
The flows configuration file is set using an environment parameter (FLOWS), which defaults to 'flows.json'. This can be changed at runtime using the following command-line flag.
$ docker run -it -p 1880:1880 -e FLOWS=my_flows.json nodered/node-red-docker
Node.js runtime arguments can be passed to the container using an environment parameter (NODE_OPTIONS). For example, to fix the heap size used by the Node.js garbage collector you would use the following command.
$ docker run -it -p 1880:1880 -e NODE_OPTIONS="--max_old_space_size=128" nodered/node-red-docker
Installing extra Node-RED nodes into an instance running with Docker can be achieved by manually installing those nodes into the container, using the cli or running npm commands within a container shell, or mounting a host directory with those nodes as a data volume.
Using the administration tool, with port forwarding on the container to the host system, extra nodes can be installed without leaving the host system.
$ npm install -g node-red-admin
$ node-red-admin install node-red-node-openwhisk
This tool assumes Node-RED is available at the following address
http://localhost:1880.
Refreshing the browser page should now reveal the newly added node in the palette.
$ docker exec -it mynodered /bin/bash
Will give a command line inside the container - where you can then run the npm install command you wish - e.g.
$ cd /data
$ npm install node-red-node-smooth
[email protected] node_modules/node-red-node-smooth
$ exit
$ docker stop mynodered
$ docker start mynodered
Refreshing the browser page should now reveal the newly added node in the palette.
Creating a new Docker image, using the public Node-RED images as the base image, allows you to install extra nodes during the build process.
This Dockerfile builds a custom Node-RED image with the flightaware module installed from NPM.
FROM nodered/node-red-docker
RUN npm install node-red-contrib-flightaware
Alternatively, you can modify the package.json in this repository and re-build the images from scratch. This will also allow you to modify the version of Node-RED that is installed. See README under docker-custom directory.
Once you have customised the Node-RED instance running with Docker, we need to ensure these modifications are not lost if the container is destroyed. Managing this user data can be handed by persisting container state into a new image or using named data volumes to handle move this data outside the container.
Modifications to files within the live container, e.g. manually adding nodes or creating flows, do not exist outside the lifetime of the container. If that container instance is destroyed, these changes will be lost.
Docker allows you to the current state of a container to a new image. This means you can persist your changes as a new image that can be shared on other systems.
$ docker commit mynodered custom-node-red
If we destroy the mynodered container, the instance can be recovered by
spawning a new container using the custom-node-red image.
Docker supports using data volumes to store persistent or shared data outside the container. Files and directories within data volumes exist outside of the lifecycle of containers, i.e. the files still exist after removing the container.
Node-RED uses the /data directory to store user configuration data.
Mounting a data volume inside the container at this directory path means user configuration data can be saved outside of the container and even shared between container instances.
Let's create a new named data volume to persist our user data and run a new container using this volume.
$ docker volume create --name node_red_user_data
$ docker volume ls
DRIVER VOLUME NAME
local node_red_user_data
$ docker run -it -p 1880:1880 -v node_red_user_data:/data --name mynodered nodered/node-red-docker
Using Node-RED to create and deploy some sample flows, we can now destroy the container and start a new instance without losing our user data.
$ docker rm mynodered
$ docker run -it -p 1880:1880 -v node_red_user_data:/data --name mynodered nodered/node-red-docker
Updating the base container image is as simple as
$ docker pull nodered/node-red-docker
$ docker stop mynodered
$ docker start mynodered
The barest minimum we need to just run Node-RED is
$ docker run -d -p 1880 nodered/node-red-docker
This will create a local running instance of a machine - that will have some docker id number and be running on a random port... to find out run
$ docker ps -a
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
4bbeb39dc8dc nodered/node-red-docker:latest "npm start" 4 seconds ago Up 4 seconds 0.0.0.0:49154->1880/tcp furious_yalow
$
You can now point a browser to the host machine on the tcp port reported back, so in the example
above browse to http://{host ip}:49154
You can link containers "internally" within the docker runtime by using the --link option.
For example I have a simple MQTT broker container available as
docker run -it --name mybroker nodered/node-red-docker
(no need to expose the port 1883 globally unless you want to... as we do magic below)
Then run nodered docker - but this time with a link parameter (name:alias)
docker run -it -p 1880:1880 --name mynodered --link mybroker:broker nodered/node-red-docker
the magic here being the --link that inserts a entry into the node-red instance
hosts file called broker that links to the mybroker instance.... but we do
expose the 1880 port so we can use an external browser to do the node-red editing.
Then a simple flow like below should work - using the alias broker we just set up a second ago.
[{"id":"190c0df7.e6f3f2","type":"mqtt-broker","broker":"broker","port":"1883","clientid":""},{"id":"37963300.c869cc","type":"mqtt in","name":"","topic":"test","broker":"190c0df7.e6f3f2","x":226,"y":244,"z":"f34f9922.0cb068","wires":[["802d92f9.7fd27"]]},{"id":"edad4162.1252c","type":"mqtt out","name":"","topic":"test","qos":"","retain":"","broker":"190c0df7.e6f3f2","x":453,"y":135,"z":"f34f9922.0cb068","wires":[]},{"id":"13d1cf31.ec2e31","type":"inject","name":"","topic":"","payload":"","payloadType":"date","repeat":"","crontab":"","once":false,"x":226,"y":157,"z":"f34f9922.0cb068","wires":[["edad4162.1252c"]]},{"id":"802d92f9.7fd27","type":"debug","name":"","active":true,"console":"false","complete":"false","x":441,"y":261,"z":"f34f9922.0cb068","wires":[]}]
This way the internal broker is not exposed outside of the docker host - of course
you may add -p 1883:1883 etc to the broker run command if you want to see it...
Here is a list of common issues users have reported with possible solutions.
If you are seeing permission denied errors opening files or accessing host devices, try running the container as the root user.
docker run -it -p 1880:1880 --name mynodered --user=root nodered/node-red-docker
References:
If you want to access a device from the host inside the container, e.g. serial port, use the following command-line flag to pass access through.
docker run -it -p 1880:1880 --name mynodered --device=/dev/ttyACM0 nodered/node-red-docker
References: node-red/node-red#15
If you want to modify the default timezone, use the TZ environment variable with the relevant timezone.
docker run -it -p 1880:1880 --name mynodered -e TZ=Europe/London nodered/node-red-docker
References: https://groups.google.com/forum/#!topic/node-red/ieo5IVFAo2o