CI/CD with Phoenix, GitHub Actions, and Gigalixir

栏目: IT技术 · 发布时间: 4年前

内容简介:I like to start my side projects by immediately configuring them with a CI pipeline and automatic deployments. I normally go with Heroku, but Heroku has some drawbacks when deploying Elixir applications.Luckily, another PaaS solution exists and is designed

I like to start my side projects by immediately configuring them with a CI pipeline and automatic deployments. I normally go with Heroku, but Heroku has some drawbacks when deploying Elixir applications.

Luckily, another PaaS solution exists and is designed specifically for Elixir, Gigalixir ! Let’s build a project called “Pipsqueak” and learn how to automatically test and deploy a new Phoenix application to Gigalixir.

Note: This article will assume that you already know how to install Elixir/Erlang and will reference tools like hub that I actually used while implementing the application that I describe below.

Create a new Phoenix project

Our first step is to create a new Phoenix application. We’ll then immediately initialize our git repository and push it up to GitHub.

I like to make the first commit right after generating the project so you don’t get lost when making your initial changes. This way, when you look at the second commit, you’ll see the first changes that the developer made.

$ mix phx.new pipsqueak && cd pipsqueak

$ git add . && git commit -m "Initial Commit"

$ hub create

$ git push -u origin master

Switch to Yarn

I like to use Yarn instead of NPM to manage my JavaScript packages, so we’ll re-install the packages using Yarn to create our yarn.lock file.

If you don’t use Yarn, you can skip this step.

$ rm assets/package-lock.json

$ (cd assets && yarn install)

$ git add . && git commit -m "Switch to Yarn"

Continuous Integration with GitHub Actions

GitHub Actions is a new product by GitHub that is used to run arbitrary workflows and CI pipelines in response to events emitted by GitHub. Let’s run our new test suite by implementing a Workflow.

Our workflow will run on every push, check that our code is formatted, and run our test suite.

Notable pieces of this workflow are:

  • We use the matrix strategy even though we are only using a single combination. This way we can use our version in our cache keys, so that we update cache will invalidate when we inevitably update our language versions.
  • We boot up our Postgres server as a service container. Make sure the user/password/db match your application’s test config.
  • We cache our Elixir and JavaScript dependencies so that we only install them if we really have to.
  • We cache our Elixir build so that the compiler only has to compile the files that have changed.
# .github/workflows/verify.yml

on: push

jobs:
  verify:
    runs-on: ubuntu-latest
    strategy:
      matrix:
        otp: [22.2.7]
        elixir: [1.10.1]

    services:
      db:
        image: postgres:12
        env:
          POSTGRES_USER: postgres
          POSTGRES_PASSWORD: postgres
          POSTGRES_DB: pipsqueak_test
        ports: ['5432:5432']
        options: --health-cmd pg_isready --health-interval 10s --health-timeout 5s --health-retries 5

    steps:
      - uses: actions/checkout@v2
      - uses: actions/setup-elixir@v1
        with:
          otp-version: ${{ matrix.otp }}
          elixir-version: ${{ matrix.elixir }}

      - name: Setup Node
        uses: actions/setup-node@v1
        with:
          node-version: 13.8.0

      - uses: actions/cache@v1
        id: deps-cache
        with:
          path: deps
          key: ${{ runner.os }}-mix-${{ hashFiles(format('{0}{1}', github.workspace, '/mix.lock')) }}

      - uses: actions/cache@v1
        id: build-cache
        with:
          path: _build
          key: ${{ runner.os }}-build-${{ matrix.otp }}-${{ matrix.elixir }}-${{ hashFiles(format('{0}{1}', github.workspace, '/mix.lock')) }}

      - name: Find yarn cache location
        id: yarn-cache
        run: echo "::set-output name=dir::$(yarn cache dir)"

      - uses: actions/cache@v1
        with:
          path: ${{ steps.yarn-cache.outputs.dir }}
          key: ${{ runner.os }}-yarn-${{ hashFiles('**/yarn.lock') }}
          restore-keys: |
            ${{ runner.os }}-yarn-

      - name: Install deps
        run: |
          mix deps.get
          (cd assets && yarn)

      - run: mix format --check-formatted
      - run: mix test

To test our workflow, let’s commit and push to GitHub.

$ git add . && git commit -m "Verify workflow"

$ git push origin master

Create a new Gigalixir app

Before proceeding with this section, please refer to the Gigalixir Getting Started Guide to create an account and install the gigalixir command line tool.

Our application will deploy using mix releases, so we need to add a config/releases.exs file. Gigalixir will detect this file and assume we are using a mix release.

The following will ensure that our server is started and URLs will be properly constructed.

# config/releases.exs

import Config

config :pipsqueak, PipsqueakWeb.Endpoint,
  server: true,
  http: [port: {:system, "PORT"}],
  url: [host: System.get_env("APP_NAME") <> ".gigalixirapp.com", port: 443]

Next, let’s configure the Elixir, Erlang, Node, and Yarn versions to be used with our deployment. We’ll use the latest version that are available at the time of writing.

Create the following files in the root of repository.

# elixir_buildpack.config

elixir_version=1.10.2
erlang_version=22.2.8

# phoenix_static_buildpack.config

node_version=13.8.0
yarn_version=1.22.0

Time to commit the changes we’ve made.

$ git add . && git commit -m "Configure for Gigalixir"

Before we can deploy our application, we’ll have to create an app and a database using the gigalixir command line tool.

$ gigalixir create

$ gigalixir pg:create --free

Now we have somewhere to deploy our code. Like Heroku, Gigalixir uses git for deployments. Running gigalixir create added a git remote named gigalixir and now we can push to it.

$ git push gigalixir master

You should now be able to visit <your app name>.gigalixirapp.com and see our stock Phoenix application!

CI/CD with Phoenix, GitHub Actions, and Gigalixir
https://pipsqueak.gigalixirapp.com

While this is quick and easy when first starting out, it can be tedious and error prone in the long run.

This is where Gigalixir falls short compared to Heroku. Heroku connects to GitHub and automatically deploys when all of your checks have passed. Checks are GitHub integrations that run some sort of test on your code and then either pass or fail.

Since we’re using GitHub Actions, we have an entire marketplace of actions at our disposal. I saw that there wasn’t an existing action for Gigalixir, so I decided to write my own!

Gigalixir Action

Check out the source code for this action here .

This action has a few features:

  • Deploy our application to Gigalixir.
  • Run our migrations upon a successful deployment.
  • If our migrations fail, it will rollback our deployment to the last version.

Running migrations requires ssh access to the deployment, so before we can get started we need to generate a new public/private key pair, upload the public key to Gigalixir, and add the private key to GitHub as a secret.

Let’s generate an SSH key pair (without a passphrase) called gigalixir_rsa in our current directory and add it to Gigalixir.

$ ssh-keygen -t rsa -b 4096 -C "Gigalixir SSH Key"

$ gigalixir account:ssh_keys:add "$(cat gigalixir_rsa.pub)"

$ cat gigalixir_rsa | pbcopy

We also copied our private key to the clipboard so that we can add it as a GitHub secret.

CI/CD with Phoenix, GitHub Actions, and Gigalixir
Adding our SSH_PRIVATE_KEY as a secret

We also need to need to add our GIGALIXIR_USERNAME and GIGALIXIR_PASSWORD as secrets in the same way we added the SSH_PRIVATE_KEY .

For now you can use the same account that you used to create the app, but in the future you should create a separate user for added security.

We can now configure the gigalixir-action to deploy our code.

# .github/workflows/verify.yml

jobs: 
  verify:
    ...

  deploy:
    # only run this job if the verify job succeeds
    needs: verify

    # only run this job if the workflow is running on the master branch
    if: github.ref == 'refs/heads/master'

    runs-on: ubuntu-latest

    steps:
      - uses: actions/checkout@v2
        
        # actions/checkout@v2 only checks out the latest commit,
        # so we need to tell it to check out the entire master branch
        with:
          ref: master
          fetch-depth: 0

      # configure the gigalixir-actions with our credentials and app name
      - uses: mhanberg/gigalixir-action@v0.1.0
        with:
          GIGALIXIR_USERNAME: ${{ secrets.GIGALIXIR_USERNAME }}
          GIGALIXIR_PASSWORD: ${{ secrets.GIGALIXIR_PASSWORD }}
          GIGALIXIR_APP: <your app name>
          SSH_PRIVATE_KEY: ${{ secrets.SSH_PRIVATE_KEY }}

Now all we have to do is commit and push our code!

$ git add . && git commit -m "Automatically deploy to Gigalixir"

What’s next?

Now there’s nothing holding you back from moving forward with your next project :smile:.

If this article has helped you deploy your latest project, please let me know on Twitter!


以上就是本文的全部内容,希望本文的内容对大家的学习或者工作能带来一定的帮助,也希望大家多多支持 码农网

查看所有标签

猜你喜欢:

本站部分资源来源于网络,本站转载出于传递更多信息之目的,版权归原作者或者来源机构所有,如转载稿涉及版权问题,请联系我们

注意力经济: 如何把大众的注意力变成生意

注意力经济: 如何把大众的注意力变成生意

吴修铭 / 中信出版集团股份有限公司 / 2018-4-1 / 69

编辑推荐 这本书由万维钢作序,并在《得到》日课中多次推荐!中文版未上市之前,中文前沿媒体就在力推这本书!关于注意力争夺战的历史和现在,作者给了权威的梳理和定位! 百年来,在争夺注意力的战场上,媒体、广告、商人、企业和大众成为博弈的主角。商人是如何在注意力争夺战中获利的?媒体是如何在改变报道形式的?广告是如何进化的?以及,营销是如何变得随处可见、无孔不入的呢?这本书讲述了令商人或企业从吸......一起来看看 《注意力经济: 如何把大众的注意力变成生意》 这本书的介绍吧!

HTML 压缩/解压工具
HTML 压缩/解压工具

在线压缩/解压 HTML 代码

图片转BASE64编码
图片转BASE64编码

在线图片转Base64编码工具

UNIX 时间戳转换
UNIX 时间戳转换

UNIX 时间戳转换