Ben Bergstein

Docker development workflow with Make and Node.js

April 16, 2020

Docker is a powerful tool that is widely used today for deployment, continuous integration (CI) and other DevOps use cases.

I increasingly turn to Docker for my development workflow. I’ve found the benefits compelling; here are a few:

✅ Consistent & replicable development environment.
✅ Simple commands to set up, start and stop your project.
✅ Language-agnostic.
✅ Easily extended to add CI & deployment at any scale.

In this tutorial, we will use the basic Next.js template to create a dockerized node.js application. It assumes basic knowledge of docker terms such as image and container, in addition to basic familiarity with node.js.

Steps

My process involves the following steps:

Let’s take as an example Dockerizing a project created with Next.js.

⭐ Create Next.js app

In order to create a dockerized application, we need an application to work with. Let’s use docker from the start, by creating our next.js app within a docker container:

Create Next.js app with yarn create

$ mkdir next-project
$ cd next-project
$ docker run -it -w /app -v "$(pwd):/app" node:latest yarn create next-app
> ✔ What is your project named? … next-project
> ✔ Pick a template › Default starter app
> Creating a new Next.js app in /next-project.
$ mv next-project app

I used next-project; use the name of your project here and elsewhere.

That was a pretty intricate docker command. Let’s take a closer look.

Command Breakdown

docker run

Run a one-off command in a Docker image.

-it

Run the command interactively.

-w /app

Use /app as the working directory.

-v "$(pwd):/app"

Mount the present working directory
(pwd) as a volume at /app.

node:latest

This is the docker image
to use; the latest node image.

yarn create next-app

This is the command to run.

mv next-project app

Move the project into the app folder.
This make dockerizing easier.

This leaves us with a brand new Next.js app in ./next-project/app folder. Because we mounted the present working directory (pwd) as a volume, the filesystem changes made by yarn create next-app are applied to our local filesystem.

Next, we’ll cover how to wrap our newly created Next.js app up in our own docker image, and how to start a container based on that image.

📜 Write Dockerfile and build image

Writing a Dockerfile will allow us to build a fully contained image of our application, ready to run on any system with Docker configured. For our Next.js application to run within docker, our Dockerfile needs to do a few things:

  • Add package.json and install dependencies.
  • Add codebase.
  • Build Next.js app.
  • Start Next.js app.

It looks like this:

Dockerfile

# ./next-project/app/Dockerfile
FROM node:latest

WORKDIR /app

COPY package.json /app
RUN yarn

COPY . /app

RUN yarn build

ENTRYPOINT ["yarn"]
CMD ["start"]

Let’s build the image:

$ cd app
$ docker build . -t next-project:latest
> Step 1/8 : FROM node:latest
> # ...
> Successfully tagged next-project:latest

Start the app

Next, let’s use the image to start a container running the app:

$ docker create --name next_project_app next-project:latest
$ docker start next_project_app
$ docker logs next_project_app
> yarn run v1.21.1
> $ next start
> > Ready on http://localhost:3000

Great! Now stop and remove the container:

$ docker rm -f next_project_app

Excellent! Still, it’s a lot to type every time we want to build the image, or start the app. Make targets will streamline these and other common tasks.

🎯 Add Make targets

Make targets will allow us to easily run common operations against our application, such as starting and stopping, and accessing logs. We’ll also use docker-compose to configure settings on the docker container, such as port, volumes, and command. Let’s start by writing the docker-compose configuration, so we have a container to start and stop.

docker-compose.yml

# ./next-project/docker/docker-compose.yml
version: '3.7'
services:
  # `app` is the name of our service
  app:
    # $DOCKER_IMAGE comes from our Makefile.
    image: ${DOCKER_IMAGE}
    # Make the port configurable. Important as many libraries default to 3000.
    ports:
      - ${PORT}:3000
    # We'll run the dev command here. For production deployment we'll change this.
    command: ["dev"]
    # Mount our codebase as a volume, so we can edit code in realtime.
    volumes:
      - ./app:/app

Then, we can create targets to build, start and stop our service. The export statements help remove repetition. For example, you can change PROJECT to the name of your project, and this will result in a docker image named based on your project’s name.

Makefile

# ./next-project/docker/Makefile
export PROJECT=next-project
export DOCKER_IMAGE=${PROJECT}:latest
export SERVICE=app
export PORT ?= 3000

define COMPOSE_CMD
docker-compose -p ${PROJECT} \
  --project-directory=.. \
  -f ./docker-compose.yml
endef

build:
	docker build ../${SERVICE} -t ${DOCKER_IMAGE}

start:
	${COMPOSE_CMD} up -d ${SERVICE}

logs:
	${COMPOSE_CMD} logs -f ${SERVICE}

stop:
	${COMPOSE_CMD} down --remove-orphans

To use these targets, run the following from the root of your project:

$ make -C docker build start logs
> make: Entering directory '/root/next-project/docker'
> docker build ../app -t next-project:latest
> Sending build context to Docker daemon  66.14MB
> Step 1/8 : FROM node:latest
> # ...
> Successfully tagged next-project:latest
> docker-compose -p next-project --project-directory=.. -f ./docker-compose.yml up -d app
> Creating network "next-project_default" with the default driver
> Creating next-project_app_1 ... done
> docker-compose -p next-project --project-directory=.. -f ./docker-compose.yml logs -f app
> Attaching to next-project_app_1
> app_1  | yarn run v1.21.1
> app_1  | $ next dev
> # ...
> app_1  | [ ready ] compiled successfully - ready on http://localhost:3000

And, voila! Our application is running in a docker container.

./next-js-started.png

In the final section, let’s examine how to access a shell within the container, which is usually done for debugging purposes or to run one-off commands.

Bonus: Add a console target

Because we mounted our codebase as a volume in the docker-compose.yml file, we can edit the code of our application as if it was not dockerized. But how do we access a console for debugging or running one-off commands?

Let’s add a simple make target for this purpose:

diff --git a/docker/Makefile b/docker/Makefile
index 1a40760..0aee168 100644
--- a/docker/Makefile
+++ b/docker/Makefile
@@ -15,6 +15,9 @@ build:
 start:
        ${COMPOSE_CMD} up -d ${SERVICE}

+console:
+       ${COMPOSE_CMD} exec ${SERVICE} /bin/bash
+
 logs:
        ${COMPOSE_CMD} logs -f ${SERVICE}

Now let’s run this target:

$ make -C docker console
> make: Entering directory '/root/next-project/docker'
> docker-compose -p next-project --project-directory=.. -f ./docker-compose.yml exec app /bin/bash
root@d38a76fb7137:/app#

We can now run commands against a shell within our node.js application’s docker container.

Conclusion

Dockerizing your application is just the beginning of the benefits of docker. We can also use docker to start additional services, such as relational databases, in-memory datastores, and even other applications required by our front-end Next.js application.

Future posts might cover how to create a backend service connected to a database with Prism.js and GraphQL.


© 2020 Benjamin Bergstein