Docker and Node.js Best Practices
Table of Contents
- Environment Variables
- Global npm dependencies
- Upgrading/downgrading Yarn
- Handling Kernel Signals
- Non-root User
- Docker Run
- node-gyp in alpine variant
NODE_ENV set to
production. This is the way you would pass in secrets and other runtime configurations to your application as well.
Global npm dependencies
If you need to install global npm dependencies, it is recommended to place those dependencies in the non-root user directory. To achieve this, add the following line to your
ENV NPM_CONFIG_PREFIX=/home/node/.npm-global ENV PATH=$PATH:/home/node/.npm-global/bin # optionally if you want to run npm global bin without specifying path
If you need to upgrade/downgrade
yarn for a local install, you can do so by issuing the following commands in your
Note that if you create some other directory which is not a descendant one from where you ran the command, you will end up using the global (dated) version. If you wish to upgrade
yarnglobally follow the instructions in the next section.
When following the local install instructions, due to duplicated yarn the image will end up being bigger.
FROM node:6 ENV YARN_VERSION 1.16.0 RUN yarn policies set-version $YARN_VERSION
FROM node:6 ENV YARN_VERSION 1.16.0 RUN curl -fSLO --compressed "https://yarnpkg.com/downloads/$YARN_VERSION/yarn-v$YARN_VERSION.tar.gz" \ && tar -xzf yarn-v$YARN_VERSION.tar.gz -C /opt/ \ && ln -snf /opt/yarn-v$YARN_VERSION/bin/yarn /usr/local/bin/yarn \ && ln -snf /opt/yarn-v$YARN_VERSION/bin/yarnpkg /usr/local/bin/yarnpkg \ && rm yarn-v$YARN_VERSION.tar.gz
If you're using an Alpine-based image,
curl won't be present, so you'll need to make sure it's installed while using it:
FROM node:6-alpine ENV YARN_VERSION 1.5.1 RUN apk add --no-cache --virtual .build-deps-yarn curl \ && curl -fSLO --compressed "https://yarnpkg.com/downloads/$YARN_VERSION/yarn-v$YARN_VERSION.tar.gz" \ && tar -xzf yarn-v$YARN_VERSION.tar.gz -C /opt/ \ && ln -snf /opt/yarn-v$YARN_VERSION/bin/yarn /usr/local/bin/yarn \ && ln -snf /opt/yarn-v$YARN_VERSION/bin/yarnpkg /usr/local/bin/yarnpkg \ && rm yarn-v$YARN_VERSION.tar.gz \ && apk del .build-deps-yarn
Handling Kernel Signals
Node.js was not designed to run as PID 1 which leads to unexpected behaviour when running inside of Docker. For example, a Node.js process running as PID 1 will not respond to
CTRL-C) and similar signals. As of Docker 1.13, you can use the
--init flag to wrap your Node.js process with a lightweight init system that properly handles running as PID 1.
docker run -it --init node
You can also include Tini directly in your Dockerfile, ensuring your process is always started with an init wrapper.
By default, Docker runs container as root which inside of the container can pose as a security issue. You would want to run the container as an unprivileged user wherever possible. The node images provide the
node user for such purpose. The Docker Image can then be run with the
node user in the following way:
Alternatively, the user can be activated in the
FROM node:6.10.3 ... # At the end, set the user to use when running this image USER node
Note that the
node user is neither a build-time nor a run-time dependency and it can be removed or altered, as long as the functionality of the application you want to add to the container does not depend on it.
If you do not want nor need the user created in this image you can remove it with the following:
# For debian based images use: RUN userdel -r node # For alpine based images use: RUN deluser --remove-home node
If you need to change the uid/gid of the user you can use:
RUN groupmod -g 999 node && usermod -u 999 -g 999 node
If you need another name for the user (ex.
RUN usermod -d /home/myapp -l myapp node
For alpine based images, you do not have
usermod, so to change the uid/gid you have to delete the previous user:
RUN deluser --remove-home node \ && addgroup -S node -g 999 \ && adduser -S -G node -u 999 node
By default, any Docker Container may consume as much of the hardware such as CPU and RAM. If you are running multiple containers on the same host you should limit how much memory they can consume.
-m "300M" --memory-swap "1G"
When creating an image, you can bypass the
start command and bake it directly into the image itself. First off this reduces the number of processes running inside of your container. Secondly it causes exit signals such as
SIGINT to be received by the Node.js process instead of npm swallowing them.
Here is an example of how you would run a default Node.JS Docker Containerized application:
$ docker run \ -e "NODE_ENV=production" \ -u "node" \ -m "300M" --memory-swap "1G" \ -w "/home/node/app" \ --name "my-nodejs-app" \ node [script]
The Docker team has provided a tool to analyze your running containers for potential security issues. You can download and run this tool from here: https://github.com/docker/docker-bench-security
Here is an example of how you would install dependencies for packages that require node-gyp support on the alpine variant:
FROM node:alpine RUN apk add --no-cache --virtual .gyp python make g++ \ && npm install [ your npm dependencies here ] \ && apk del .gyp
And Here's a multistage build example
FROM node:alpine as builder ## Install build toolchain, install node deps and compile native add-ons RUN apk add --no-cache python make g++ RUN npm install [ your npm dependencies here ] FROM node:alpine as app ## Copy built node modules and binaries without including the toolchain COPY --from=builder node_modules .