You could use this example app as a base for your new project or as a guide to Dockerize your existing Phoenix app.
The example app is minimal but it wires up a number of things you might use in a real world Phoenix app, but at the same time it's not loaded up with a million personal opinions.
For the Docker bits, everything included is an accumulation of Docker best practices based on building and deploying dozens of assorted Dockerized web apps since late 2014.
This app is using Phoenix 1.7.14 and Elixir 1.17.3. The screenshot doesn't get updated every time I bump the versions:
- Tech stack
- Notable opinions and extensions
- Running this app
- Files of interest
- Running a script to automate renaming the project
- Updating dependencies
- See a way to improve something?
- Additional resources
- About the author
If you don't like some of these choices that's no problem, you can swap them out for something else on your own.
Picking a JS library is a very app specific decision because it depends on which library you like and it also depends on if your app is going to be mostly HEEx templates with sprinkles of JS or an API back-end.
This isn't an exhaustive list but here's a few reasonable choices depending on how you're building your app:
- https://hotwired.dev/
- https://htmx.org/
- https://github.com/alpinejs/alpine
- https://vuejs.org/
- https://reactjs.org/
- https://jquery.com/
Picking a JS library is a very app specific decision because it depends on which library you like and it also depends on if your app is going to be mostly Jinja templates with sprinkles of JS or an API back-end.
Personally I'm going to be using LiveView with StimulusJS in most newer projects.
Phoenix is a healthy mix of being both opinionated and unopinionated but I find in most apps I'm adding the same things over and over. Here's a few note worthy additions and changes.
- Mix and releases:
- When
MIX_ENV
is set todev
then the server will be run with mix - When
MIX_ENV
is set toprod
a release will be built and used to run your server - All of this is automated through a multi-stage Dockerfile and build ARGs
- When
- Static analysis and testing:
- credo helps create more consistent code
- excoveralls for getting detailed test coverge reports
- Controllers:
- Add
page
controller to render a home page - Add
up
controller with 2 health check related actions
- Add
- Config:
- Extract a bunch of configuration settings into environment variables
- The
.env
file stores all environment variables - Put all config settings with environment variables in
config/runtime.exs
(works with releases)
- Phoenix defaults that are changed:
- Cookies are encrypted
- A
lib/hello/release.ex
module has been added to run migrations with releases .formatter.exs
has been configured to use 80 character line lengths- An empty
.iex.exs
file has been created so you can customize your IEx sessions however you see fit - CSS, JS and image links are referenced with
url(~p"...")
to support optionally using a CDN
- Front-end assets:
- Custom
502.html
andmaintenance.html
pages - Generate favicons using modern best practices
- Custom
Besides the Elixir / Phoenix app itself:
- Docker support has been added which would be any files having
*docker*
in its name - GitHub Actions have been set up
You'll need to have Docker installed. It's available on Windows, macOS and most distros of Linux. If you're new to Docker and want to learn it in detail check out the additional resources links near the bottom of this README.
You'll also need to enable Docker Compose v2 support if you're using Docker Desktop. On native Linux without Docker Desktop you can install it as a plugin to Docker. It's been generally available for a while now and is stable. This project uses specific Docker Compose v2 features that only work with Docker Compose v2 2.20.2+.
If you're using Windows, it will be expected that you're following along inside of WSL or WSL 2. That's because we're going to be running shell commands. You can always modify these commands for PowerShell if you want.
git clone https://github.com/nickjj/docker-phoenix-example hellophoenix
cd hellophoenix
# Optionally checkout a specific tag, such as: git checkout 0.9.0
cp .env.example .env
The first time you run this it's going to take 5-10 minutes depending on your internet connection speed and computer's hardware specs. That's because it's going to download a few Docker images and build the Elixir + Yarn dependencies.
docker compose up --build
Now that everything is built and running we can treat it like any other Phoenix app.
Did you receive a depends_on
"Additional property required is not allowed"
error? Please update to at least Docker Compose v2.20.2+ or Docker Desktop
4.22.0+.
Did you receive an error about a port being in use? Chances are it's because
something on your machine is already running on port 8000. Check out the docs
in the .env
file for the DOCKER_WEB_PORT_FORWARD
variable to fix this.
Did you receive a permission denied error? Chances are you're running native
Linux and your uid:gid
aren't 1000:1000
(you can verify this by running
id
). Check out the docs in the .env
file to customize the UID
and GID
variables to fix this.
Visit http://localhost:8000 in your favorite browser.
# You should get a report back saying there's no issues.
./run lint
We'll go over that ./run script in a bit!
# You can check to see if any code is not formatted correctly.
./run format:check
# You can also auto-fix any malformed code (both commands call mix format).
./run format
# You should see all passing tests. Warnings are typically ok.
./run test
# You can also get test coverage which uses coveralls.
./run test:coverage
./run test:coverage:details
# Stop the containers and remove a few Docker related resources associated to this project.
docker compose down
You can start things up again with docker compose up
and unlike the first
time it should only take seconds.
I recommend checking out most files and searching the code base for TODO:
,
but please review the .env
and run
files before diving into the rest of the
code and customizing it. Also, you should hold off on changing anything until
we cover how to customize this example app's name with an automated script
(coming up next in the docs).
This file is ignored from version control so it will never be commit. There's a number of environment variables defined here that control certain options and behavior of the application. Everything is documented there.
Feel free to add new variables as needed. This is where you should put all of your secrets as well as configuration that might change depending on your environment (specific dev boxes, CI, production, etc.).
You can run ./run
to get a list of commands and each command has
documentation in the run
file itself.
It's a shell script that has a number of functions defined to help you interact
with this project. It's basically a Makefile
except with less
limitations.
For example as a shell script it allows us to pass any arguments to another
program.
This comes in handy to run various Docker commands because sometimes these commands can be a bit long to type. Feel free to add as many convenience functions as you want. This file's purpose is to make your experience better!
If you get tired of typing ./run
you can always create a shell alias with
alias run=./run
in your ~/.bash_aliases
or equivalent file. Then you'll be
able to run run
instead of ./run
.
The app is named hello
right now but chances are your app will be a different
name. Since the app is already created we'll need to do a find / replace on a
few variants of the string "hello" and update a few Docker related resources.
And by we I mean I created a zero dependency shell script that does all of the heavy lifting for you. All you have to do is run the script below.
# The script takes 2 arguments.
#
# The first one is the lower case version of your app's name, such as myapp or
# my_app depending on your preference.
#
# The second one is used for your app's module name. For example if you used
# myapp or my_app for the first argument you would want to use MyApp here.
bin/rename-project myapp MyApp
The bin/rename-project script is going to:
- Remove any Docker resources for your current project
- Perform a number of find / replace actions
- Optionally initialize a new git repo for you
Afterwards you can delete this script because its only purpose is to assist in helping you change this project's name without depending on any complicated project generator tools or 3rd party dependencies.
If you're not comfy running the script or it doesn't work for whatever reasons you can check it out and perform the actions manually. It's mostly running a find / replace across files and then renaming a few directories and files.
This won't take as long as before because Docker can re-use most things.
docker compose up --build
It's always a good idea to make sure things are in a working state before adding custom changes.
# You can run this from the same terminal as before.
./run lint
./run format
./run test
If everything passes now you can optionally git add -A && git commit -m "Initial commit"
and start customizing your app. Alternatively you can wait
until you develop more of your app before committing anything. It's up to you!
You'll probably want to create a fresh CHANGELOG.md
file for your project. I
like following the style guide at https://keepachangelog.com/ but feel free
to use whichever style you prefer.
Since this project is MIT licensed you should keep my name and email address in
the LICENSE
file to adhere to that license's agreement, but you can also add
your name and email on a new line.
If you happen to base your app off this example app or write about any of the code in this project it would be rad if you could credit this repo by linking to it. If you want to reference me directly please link to my site at https://nickjanetakis.com. You don't have to do this, but it would be very much appreciated!
Let's say you've customized your app and it's time to make a change to your
mix.exs
or package.json
file.
Without Docker you'd normally run mix deps.get
or yarn install
. With Docker
it's basically the same thing and since these commands are in our Dockerfile
we can get away with doing a docker compose build
but don't run that just
yet.
You can run ./run mix:outdated
or ./run yarn:outdated
to get a list of
outdated dependencies based on what you currently have installed. Once you've
figured out what you want to update, go make those updates in your mix.exs
and / or assets/package.json
file.
Then to update your dependencies you can run ./run mix:install
or ./run yarn:install
. That'll make sure any lock files get copied from Docker's image
(thanks to volumes) into your code repo and now you can commit those files to
version control like usual.
You can check out the run file to see what these commands do in more detail.
You'll want to run docker compose build
since it will use any existing lock
files if they exist. You can also check out the complete CI test pipeline in
the run file
under the ci:test
function.
This is usually a non-issue since you'll be pulling down pre-built images from
a Docker registry but if you decide to build your Docker images directly on
your server you could run docker compose build
as part of your deploy
pipeline.
If you see anything that could be improved please open an issue or start a PR. Any help is much appreciated!
Now that you have your app ready to go, it's time to build something cool! If you want to learn more about Docker, Phoenix and deploying a Phoenix app here's a couple of free and paid resources. There's Google too!
- https://diveintodocker.com is a course I created which goes over the Docker and Docker Compose fundamentals
I'm creating an in-depth course related to deploying Dockerized web apps. If you want to get notified when it launches with a discount and potentially get free videos while the course is being developed then sign up here to get notified.
- Nick Janetakis | https://nickjanetakis.com | @nickjanetakis
I'm a self taught developer and have been freelancing for the last ~20 years. You can read about everything I've learned along the way on my site at https://nickjanetakis.com.
There's hundreds of blog posts / videos and a couple of video courses on web development and deployment topics. I also have a podcast where I talk with folks about running web apps in production.