Skip to content
This repository has been archived by the owner on Dec 9, 2022. It is now read-only.

machine-learning-apps/gpr-docker-publish

Use this GitHub action with your project
Add this Action to an existing workflow or create a new one
View on Marketplace

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

67 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Actions Status License: MIT

This Action Publishes Docker Images to the GitHub Package Registry.

Background On The GitHub Package Registry (GPR):

From the docs:

The GitHub Package Registry allows you to develop your code and host your packages in one place. GitHub uses the README in your repository to generate the package's description, and you can edit it to reflect details about the package or installation process. GitHub adds metadata for each package version that includes links to the author, repository, commit SHA, version tags, and date.

The docs also contain relevant such as how to authenticate and naming conventions. Some noteable items about publishing Docker Images on GPR:

  • Docker Images are tied to a repository.

  • All images are named with the following nomenclature:

    docker.pkg.github.com/{OWNER}/{REPOSITORY}/{IMAGE_NAME}:{TAG}

OWNER and REPOSITORY refer to a unique repository on GitHub, such as tensorflow/tensorflow.

Automatic Tagging

This Action will automatically tag each image as follows:

{Image_Name}:{shortSHA}

Where:

  • Image_Name is provided by the user as an input.
  • shortSHA is the first 12 characters of the GitHub SHA that triggered the action.

Usage

Example Workflow That Uses This Action:

name: Publish Docker image to GitHub Package Registry
on: push
jobs:
  build:
    runs-on: ubuntu-latest 
    steps:

    - name: Copy Repo Files
      uses: actions/checkout@master

     #This Action Emits 2 Variables, IMAGE_SHA_NAME and IMAGE_URL 
     #which you can reference in subsequent steps
    - name: Publish Docker Image to GPR
      uses: machine-learning-apps/gpr-docker-publish@master
      id: docker
      with:
        IMAGE_NAME: 'test-docker-action'
        TAG: 'my-optional-tag-name'
        DOCKERFILE_PATH: 'argo/gpu.Dockerfile'
        BUILD_CONTEXT: 'argo/'
      env:
        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

    # This second step is illustrative and shows how to reference the 
    # output variables.  This is completely optional.
    - name: Show outputs of previous step
      run: |
        echo "The name:tag of the Docker Image is: $VAR1"
        echo "The docker image is hosted at $VAR2"
      env:
        VAR1: ${{ steps.docker.outputs.IMAGE_SHA_NAME }}
        VAR2: ${{ steps.docker.outputs.IMAGE_URL }}

Mandatory Inputs

  1. IMAGE_NAME is the name of the image you would like to push
  2. DOCKERFILE_PATH: The full path (including the filename) relative to the root of the repository that contains the Dockerfile that specifies your build.
  3. BUILD_CONTEXT: The directory for the build context. See these docs for more information on the definition of build context.

Optional Inputs

  1. cache: if value is true, attempts to use the last pushed image as a cache. Default value is false.
  2. tag: a custom tag you wish to assign to the image.
  3. branch_tag: if value is true, will tag the image with the current branch's name. The default value is false. If a value for the input tag is provided, this input is ignored.

Outputs

You can reference the outputs of an action using expression syntax, as illustrated in the Example Pipeline above.

  1. IMAGE_SHA_NAME: This is the {Image_Name}:{shortSHA} as described above.
  2. IMAGE_URL: This is the URL on GitHub where you can view your hosted Docker images. This will always be located at https://github.com/{OWNER}/{REPOSITORY}/packages in reference to the repository where the action was called.

These outputs are merely provided as convenience incase you want to use these values in subsequent steps.