Skip to content
Branch: master
Find file Copy path
Find file Copy path
6 contributors

Users who have contributed to this file

@RaymondMouthaan @dceejay @jthomas @knolleary @greenkeeper @cpswan
519 lines (373 sloc) 25.5 KB

Node-RED Docker

Greenkeeper badge Build Status DockerHub Pull DockerHub Stars

This project describes some of the many ways Node-RED can be run under Docker and has support for multiple architectures (amd64, arm32v6, arm32v7, arm64v8, i386 and s390x). Some basic familiarity with Docker and the Docker Command Line is assumed.

As of Node-RED 1.0 this project provides the build for the nodered/node-red container on Docker Hub. Note: the Docker Hub name has changed to nodered/node-red.

Previous 0.20.x versions are still available at

Quick Start

To run in Docker in its simplest form just run:

    docker run -it -p 1880:1880 --name mynodered nodered/node-red

Let's dissect that command:

    docker run              - run this container, initially building locally if necessary
    -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        - the image to base it on - currently Node-RED v1.0.2

Running that command should give a terminal window with a running instance of Node-RED.

    Welcome to Node-RED

    03 Oct 12:57:10 - [info] Node-RED version: v1.0.2
    03 Oct 12:57:10 - [info] Node.js  version: v10.16.3
    03 Oct 12:57:10 - [info] Linux 4.9.184-linuxkit x64 LE
    03 Oct 12:57:11 - [info] Loading palette nodes
    03 Oct 12:57:16 - [info] Settings file  : /data/settings.js
    03 Oct 12:57:16 - [info] Context store  : 'default' [module=memory]
    03 Oct 12:57:16 - [info] User directory : /data
    03 Oct 12:57:16 - [warn] Projects disabled : editorTheme.projects.enabled=false
    03 Oct 12:57:16 - [info] Flows file     : /data/flows.json
    03 Oct 12:57:16 - [info] Creating new flow file
    03 Oct 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.

    03 Oct 12:57:17 - [info] Starting flows
    03 Oct 12:57:17 - [info] Started flows
    03 Oct 12:57:17 - [info] Server now running at


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 (mynodered) 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

Image Variations

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 or minimal.
    • none : is the default and has Python 2 & Python 3 + devtools installed
    • minimal : has no Python installed and no devtools installed
  • <architecture> is the architecture of the Docker host system, can be either amd64, arm32v6, arm32v7, arm64, s390x or i386.

The minimal versions (without python and build tools) are not able to install nodes that require any locally compiled native code.

For example - to run the latest minimal version, you would run

docker run -it -p 1880:1880 --name mynodered nodered/node-red:latest-minimal

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 variety of provided Node-RED images.

Tag Node Arch Python Dev Base Image
1.0.2-10-amd64 10 amd64 2.x 3.x yes amd64/node:10-alpine
1.0.2-10-arm32v6 10 arm32v6 2.x 3.x yes arm32v6/node:10-alpine
1.0.2-10-arm32v7 10 arm32v7 2.x 3.x yes arm32v7/node:10-alpine
1.0.2-10-arm64v8 10 arm64v8 2.x 3.x yes arm64v8/node:10-alpine
1.0.2-10-s390x 10 s390x 2.x 3.x yes s390x/node:10-alpine
1.0.2-10-i386 10 i386 2.x 3.x yes i386/node:10-alpine
1.0.2-10-minimal-amd64 10 amd64 no no amd64/node:10-alpine
1.0.2-10-minimal-arm32v6 10 arm32v6 no no arm32v6/node:10-alpine
1.0.2-10-minimal-arm32v7 10 arm32v7 no no arm32v7/node:10-alpine
1.0.2-10-minimal-arm64v8 10 arm64v8 no no arm64v8/node:10-alpine
1.0.2-10-minimal-s390x 10 s390x no no s390x/node:10-alpine
1.0.2-10-minimal-i386 10 i386 no no i386/node:10-alpine
Tag Node Arch Python Dev Base Image
1.0.2-12-amd64 12 amd64 2.x 3.x yes amd64/node:12-alpine
1.0.2-12-arm32v6 12 arm32v6 2.x 3.x yes arm32v6/node:12-alpine
1.0.2-12-arm32v7 12 arm32v7 2.x 3.x yes arm32v7/node:12-alpine
1.0.2-12-arm64v8 12 arm64v8 2.x 3.x yes arm64v8/node:12-alpine
1.0.2-12-s390x 12 s390x 2.x 3.x yes s390x/node:12-alpine
1.0.2-12-i386 12 i386 2.x 3.x yes i386/node:12-alpine
1.0.2-12-minimal-amd64 12 amd64 no no amd64/node:12-alpine
1.0.2-12-minimal-arm32v6 12 arm32v6 no no arm32v6/node:12-alpine
1.0.2-12-minimal-arm32v7 12 arm32v7 no no arm32v7/node:12-alpine
1.0.2-12-minimal-arm64v8 12 arm64v8 no no arm64v8/node:12-alpine
1.0.2-12-minimal-s390x 12 s390x no no s390x/node:12-alpine
1.0.2-12-minimal-i386 12 i386 no no i386/node:12-alpine
  • All images have bash, tzdata, nano, curl git and openssl tools pre-installed to support Node-RED's Projects feature.

Manifest Lists

The following table shows the provided Manifest Lists.

Tag Node-RED Base Image
latest, 1.0.2, nodered/node-red:1.0.2-10-amd64
latest-10, 1.0.2-10 nodered/node-red:1.0.2-10-arm32v6
latest-minimal, 1.0.2-minimal, nodered/node-red:1.0.2-10-amd64-minimal
latest-10-minimal, 1.0.2-10-minimal nodered/node-red:1.0.2-10-arm32v6-minimal
Tag Node-RED Base Image
latest-12, 1.0.2-12 nodered/node-red:1.0.2-12-amd64
latest-12-minimal, 1.0.2-12-minimal nodered/node-red:1.0.2-12-amd64-minimal

With the support of Docker manifest list, there is no need to explicitly 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. If it does, docker pulls the matching image for it.

Therefore 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 run the following command to pull the image (tagged by 1.0.2-10-arm32v7), and run the container.

docker run -it -p 1880:1880 --name mynodered nodered/node-red: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 the matching tag (1.0.2-10-amd64).

This gives the advantage that you don't need to know/specify which architecture you are running on and makes docker run commands and docker compose files more flexible and exchangeable across systems.

Note: Currently there is a bug in Docker's architecture detection that fails for arm32v6 - eg Raspberry Pi Zero or 1. For these devices you currently need to specify the full image tag, for example:

docker run -it -p 1880:1880 --name mynodered nodered/node-red:1.0.2-10-minimal-arm32v6

Raspberry PI - native GPIO support

v1.0 - BREAKING: Native GPIO support for Raspberry PI has been dropped
The replacement for native GPIO is node-red-node-pi-gpiod.

Disadvantages of the native GPIO support are:

  • Your Docker container needs to be deployed on the same Docker node/host on which you want to control the gpio.
  • Gain access to /dev/mem of your Docker node/host
  • privileged=true is not supported for docker stack command

node-red-node-pi-gpiod fixes all these disadvantages. With node-red-node-pi-gpiod it is possible to interact with gpio of multiple Raspberry Pi's from a single Node-RED container, and for multiple containers to access different gpio on the same Pi.

Quick Migration steps to node-red-node-pi-gpiod

  1. Install node-red-node-pi-gpiod through the Node-RED palette
  2. Install and run PiGPIOd daemon on the host Pi.
  3. Replace all native gpio nodes with pi gpiod nodes.
  4. Configure pi gpiod nodes to connect to PiGPIOd daemon. Often the host machine will have an IP port 8888 - but not always. You can use docker exec -it mynodered ip route show default | awk '/default/ {print $3}' to check.

For detailed install instruction please refer to the node-red-node-pi-gpiod README

Note: There is a contributed gpiod project that runs the gpiod in its own container rather than on the host if required.

Managing User Data

Once you have Node-RED running with Docker, we need to ensure any added nodes or flows are not lost if the container is destroyed. This user data can be persisted by mounting a data directory to a volume outside the container. This can either be done using a bind mount or a named data volume.

Node-RED uses the /data directory inside the container to store user configuration data.

Using a Host Directory for Persistence (Bind Mount)

To save your Node-RED user directory inside the container to a host directory outside the container, you can use the command below. To allow access to this host directory, the node-red user (default uid=1000) inside the container must have the same uid as the owner of the host directory.

docker run -it -p 1880:1880 -v /home/pi/.node-red:/data --name mynodered nodered/node-red

In this example the host /home/pi/.node-red directory is bound to the container /data directory.

Note: Users migrating from version 0.20 to 1.0 will need to ensure that any existing /data directory has the correct ownership. As of 1.0 this needs to be 1000:1000. This can be forced by the command sudo chown -R 1000:1000 path/to/your/node-red/data

See the wiki for detailed information on permissions.

Using Named Data Volumes

Docker also supports using named data volumes to store persistent or shared data outside the container.

To 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

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


As the /data is now preserved outside of the container, updating the base container image is now as simple as

    $ docker pull nodered/node-red
    $ docker stop mynodered
    $ docker start mynodered

Docker Stack / Docker Compose

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 or Compose
# docker stack deploy node-red --compose-file docker-compose-node-red.yml
# docker-compose -f docker-compose-node-red.yml -p myNoderedProject up
version: "3.7"

    image: nodered/node-red:latest
      - TZ=Europe/Amsterdam
      - "1880:1880"
      - node-red-net
      - ~/node-red/data:/data


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 /data dir inside the container to the users local node-red/data directory. The node-red/data directory must exist prior to starting the container.

Project Layout

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": "^1.0.2",           <-- set the version of Node-RED here
    "node-red-dashboard": "*"        <-- 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

Note: If you set -e FLOWS="" then the flow file can be set via the flowFile property in the settings.js file.

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

Other useful environment variables include

  • -e NODE_RED_ENABLE_SAFE_MODE=false # setting to true starts Node-RED in safe (not running) mode
  • -e NODE_RED_ENABLE_PROJECTS=false # setting to true starts Node-RED with the projects feature enabled

Node-RED Admin Tool

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.

Container Shell

    $ 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
    $ exit
    $ docker stop mynodered
    $ docker start mynodered

Refreshing the browser page should now reveal the newly added node in the palette.

Building Custom Image

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
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 in the docker-custom directory.

Running headless

The barest minimum we need to just run Node-RED is

$ docker run -d -p 1880:1880 nodered/node-red

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
CONTAINER ID        IMAGE                            COMMAND             CREATED             STATUS                     PORTS                     NAMES
4bbeb39dc8dc        nodered/node-red:latest          "npm start"         4 seconds ago       Up 4 seconds     >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

Linking Containers

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 eclipse-mosquitto

(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

the magic here being the --link that inserts a entry into the node-red instance hosts file called broker that links to the external 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...

Docker-Compose linking example

Another way to link containers is by using docker-compose. The following docker-compose.yml file creates a Node-RED instance, and a local MQTT broker instance. In the Node-RED flow the broker can be addressed simply as broker at its default port 1883.

version: "3.7"

    image: nodered/node-red
    restart: unless-stopped
      - /home/pi/.node-red:/data
      - 1880:1880

    image: eclipse-mosquitto
    restart: unless-stopped

Common Issues and Hints

Here is a list of common issues users have reported with possible solutions.

User Permission Errors

See the wiki for detailed information on permissions.

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 -u root nodered/node-red


Accessing Host Devices

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


Setting Timezone

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


You can’t perform that action at this time.