Dockerized Node.js local development environment

Isolate your Node.js processes.

This could be used for running a local dev server like Vite, Next.js or Nest.js.

Reasoning

Why do this?

Docker, especially in rootless mode, allows for:

1. Greater security

   • Thousands of those NPM dependencies are now prevented from getting any glimpse of the host.

2. Improved reproducibility

   • Platform compatibility issues like environment variables on Windows vs UNIX.

   • Developers working on the app get the exact same result every time, no matter if they are running Windows, macOS or Linux.

Setup

Ensure Docker Engine and Docker Compose are set up. Refer to platform-specific instructions: https://docs.docker.com/engine/install

With Docker Desktop (GUI) your mileage may vary.

To verify, the following commands should print:

docker -v
docker compose version

Optionally, make the ./dev convenience shell script runnable:

chmod +x ./dev

This will allow us to avoid typing the verbose docker compose yada yada yada every time we want to spin up our dev environment.

For the first run, we must enter shell in order to first set up NPM/PNPM/Yarn/Bun (e.g. install vite as a dep):

./dev bash

Environment variables

Create an .env or remove env_file in docker-compose.yml.

Alternatively, env vars can be hardcoded (or otherwise piped in):

environment:
  - DEBUG=${DEBUG}

[!NOTE] Variables via docker-compose will be available in the environment of the container, i.e. OS level.
This is identical to how env vars would be available during a CI/CD build.
For development, prefer .env loaded via framework tooling with auto-refresh, e.g. Vite dev server.

See 💡 Docker Docs

Port mapping

Specify host:container ports in docker-compose.yml.

If our node.js process inside the container runs on 5173 but we want to avoid clashing with something else on our host machine, we could map it like so: 5174:5173.

See 💡 Docker Docs

Usage

[!IMPORTANT] Ensure the ./dev script is chmoded.

Useful docker commands

# List
docker ps -a
docker volume ls
docker image ls

# Clean
docker builder prune --all --force
docker system prune --all --volumes --force
# Caution: deletes volumes permanently
docker compose down --volumes

Isolating Node.js further

The repository root . (as in, current dir) is mounted as /app within the container using WORKDIR.

This means that all files in the repository root are available to any node.js process. For example,.env.prod or other secrets.

[!TIP]
WORKDIR mounts everything as a filesystem volume, granting us real-time two-way synchronization. Useful for actual development work.
COPY, in contrast, copies files over once at container build-time and respects .dockerignore.

To completely isolate a node.js app, we can mount a subdirectory instead of root .:

volumes:
  - ./app:/app # host:container

We then place package.json under app/ along with the rest of our application-specific sourceode.

Using with PNPM

Want to use PNPM or another package manager like Yarn or Bun?

Modify Dockerfile along these lines:

# ...
WORKDIR /app

ENV COREPACK_ENABLE_DOWNLOAD_PROMPT=0
ENV PNPM_HOME=/usr/local/share/.pnpm-store
ENV PATH=$PNPM_HOME:$PATH

RUN mkdir -p $PNPM_HOME

RUN npm install -g npm@latest corepack@latest
RUN corepack enable pnpm
RUN corepack use pnpm@latest

Then, ensure that docker-compose.yml knows what to run by default:

command: >
  bash -c "pnpm i && pnpm dev:docker"