Giter Club home page Giter Club logo

docker-rails-dev's Introduction

Build images

DockerRailsDev

Building Docker images usually takes a long time. This repo contains base images with preinstalled dependencies for Ruby on Rails, so building a production image will be 2-3 times faster.

What?

When using the official Ruby image, building a Docker image for a typical Rails application requires lots of time for installing dependencies - mainly OS packages, Ruby gems, Ruby gems with native extensions (Nokogiri etc.) and Node modules. This is required every time the app needs to be deployed to production.

I was looking for a way to reduce this time, so I created base images that contain most of the dependencies used in my applications.

And while I'm at it, I also moved as much as possible from the app-specific Dockerfile into the base image by using ONBUILD triggers. This makes the Dockerfile in my apps small and simple.

Performance

I compared building times using a typical Rails application. This is the result on my local machine:

  • Based on official Ruby image: 4:50 min
  • Based on DockerRailsBase: 1:57 min

As you can see, using DockerRailsDev is more than 2 times faster compared to the official Ruby image. It saves nearly 3min on every build.

Note: Before I started timing, the base image was not available on my machine, so it was downloaded first, which took some time. If the base image is already available, the building time is only 1:18min (3 times faster).

Requirements

This repo is based on the following assumptions:

  • Your Docker host is compatible with Alpine Linux 3.18, which requires Docker 20.10.0 or later
  • Your app is compatible with Ruby 3.2 for Alpine Linux
  • Your app uses Ruby on Rails 6.0, 6.1 or 7.0
  • Your app uses PostgreSQL, SQLite or MySQL/MariaDB
  • Your app installs Node modules with Yarn
  • Your app bundles JavaScript with rails assets:precompile. This works with Vite Ruby, Webpacker, Asset pipeline (Sprockets) and others. This project does not support API-only applications, as there is no JavaScript to compile there.

If your project differs from this, I suggest to fork this project and create your own base image.

How?

It uses multi-stage building to build a very small production image. There are two Dockerfiles in this repo, one for the first stage (called Builder) and one for the resulting stage (called Final).

Builder stage

The Builder stage installs Ruby gems and Node modules. It also includes Git, Node.js and some build tools - all we need to compile assets.

  • Based on ruby:3.2.2-alpine
  • Adds packages needed for installing gems and compiling assets: Git, Node.js, Yarn, PostgreSQL client and build tools
  • Adds some default Ruby gems (Rails 7.0 etc., see Gemfile)
  • Adds some default Node modules (Turbo, Stimulus, TailwindCSS etc., see package.json)
  • Via ONBUILD triggers it installs missing gems and Node modules, then compiles the assets

See Builder/Dockerfile

Final stage

The Final stage builds the production image, which includes just the bare minimum.

  • Based on ruby:3.2.2-alpine
  • Adds packages needed for production: postgresql-client, tzdata, file
  • Via ONBUILD triggers it mainly copies the app and gems from the Builder stage

See Final/Dockerfile

Staying up-to-date

Using Dependabot, every updated Ruby gem or Node module results in an updated image.

How to use for your Rails application

Building the Docker image

Add this Dockerfile to your application:

FROM krmbzds/rails-dev-builder:3.2.2-alpine AS Builder
FROM krmbzds/rails-dev-final:3.2.2-alpine
USER app
# Optional: Enable YJIT
# ENV RUBY_YJIT_ENABLE=1
CMD ["bundle", "exec", "puma", "-C", "config/puma.rb"]

Yes, this is the complete Dockerfile of your Rails app. It's so simple because the work is done by ONBUILD triggers.

Now build the image:

$ docker build .

Building the Docker image with BuildKit

BuildKit requires a little workaround to trigger the ONBUILD statements. Add a COPY statement to the Dockerfile:

FROM krmbzds/rails-dev-builder:3.2.2-alpine AS Builder
FROM krmbzds/rails-dev-final:3.2.2-alpine

# Workaround to trigger Builder's ONBUILDs to finish:
COPY --from=Builder /etc/alpine-release /tmp/dummy

USER app
# Optional: Enable YJIT
# ENV RUBY_YJIT_ENABLE=1
CMD ["bundle", "exec", "puma", "-C", "config/puma.rb"]

Now you can build the image with BuildKit:

docker buildx build .

You can use private npm/Yarn packages by mounting the config file:

docker buildx build --secret id=npmrc,src=$HOME/.npmrc .

or

docker buildx build --secret id=yarnrc,src=$HOME/.yarnrc.yml .

In a similar way you can provide a configuration file for Bundler:

docker buildx build --secret id=bundleconfig,src=$HOME/.bundle/config .

Continuous integration (CI)

Example to build the application's image with GitHub Actions and push it to the GitHub Container Registry:

deploy:
  runs-on: ubuntu-latest

  steps:
    - uses: actions/checkout@v2
      with:
        fetch-depth: 0

    - name: Fetch tag annotations
      # https://github.com/actions/checkout/issues/290
      run: git fetch --tags --force

    - name: Login to GitHub Container Registry
      uses: docker/login-action@v1
      with:
        registry: ghcr.io
        username: ${{ github.repository_owner }}
        password: ${{ secrets.GITHUB_TOKEN }}

    - name: Build the image
      run: |
        export COMMIT_TIME=$(git show -s --format=%cI ${GITHUB_SHA})
        export COMMIT_SHA=${GITHUB_SHA}
        export COMMIT_VERSION=$(git describe)
        docker buildx build --build-arg COMMIT_TIME --build-arg COMMIT_SHA --build-arg COMMIT_VERSION -t ghcr.io/user/repo:latest .

    - name: Push the image
      run: docker push ghcr.io/user/repo:latest

Available Docker images

Both Docker images (Builder and Final) are regularly published at DockerHub and tagged with the current Ruby version:

Beware: The published images are not immutable. When a dependency (e.g. Ruby gem) is updated, the images will be republished using the same tag.

When a new Ruby version comes out, a new tag is introduced and the images will be published using this tag and the former images will not be updated anymore. Here is a list of the tags that have been used in this repo so far:

Ruby version Tag First published
3.2.2 3.2.2-alpine 2023-03-31
3.2.1 3.2.1-alpine 2023-02-10
3.2.0 3.2.0-alpine 2023-01-13
3.1.3 3.1.3-alpine 2022-11-26
3.1.2 3.1.2-alpine 2022-04-13
3.1.1 3.1.1-alpine 2022-02-19
3.1.0 3.1.0-alpine 2022-01-08
3.0.3 3.0.3-alpine 2021-11-24
3.0.2 3.0.2-alpine 2021-07-08
3.0.1 3.0.1-alpine 2021-04-06
3.0.0 3.0.0-alpine 2021-02-15
2.7.2 2.7.2-alpine 2020-10-10
2.7.1 2.7.1-alpine 2020-05-20
2.6.6 - 2020-04-01
2.6.5 - 2020-01-24

The latest Docker images are also tagged as latest. However, it is not recommended to use this tag in your Rails application, because updating an app to a new Ruby version usually requires some extra work.

FAQ

Why not simply use layer caching?

Docker supports layer caching, so for building images it performs just the needed steps: If there is a layer from a former build and nothing has changed, it will be used. But for dependencies, this means: If a single Ruby gem in the application was updated or added, the step with bundle install is run again, so all gems will be installed again.

Using a prebuilt image improves installing dependencies a lot, because only the different/updated dependencies will be installed - all existing ones will be reused.

What if my app requires slightly different dependencies?

This doesn't matter:

  • A missing Alpine package can be installed with apk add inside your app's Dockerfile.
  • A missing Node module (or version) will be installed with rails assets:precompile via the ONBUILD trigger.
  • A missing Ruby gem (or version) will be installed with bundle install via the ONBUILD trigger.

There are gems included that my app doesn't need. Will they bloat the resulting image?

No. In the build stage there is a bundle clean --force, which uninstalls all gems not referenced in the app's Gemfile.

docker-rails-dev's People

Contributors

dependabot[bot] avatar dependabot-preview[bot] avatar ledermann avatar juice10 avatar deepakinseattle avatar krmbzds avatar

Watchers

 avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.