Initial taskcat GitHub Action MVP about action-taskcat

[ShahradR]

By including a sample workflow under .github/workflows/main.yaml, we are able to simulate the action's execution using act.

To run the action without pushing code to a repository, set uses: ./ in the workflow definition, as shown below.

---
on: [push]

jobs:
  taskcat:
    runs-on: ubuntu-latest
    name: Validate CloudFormation templates
    steps:
      - name: Run taskcat
        id: taskcat
        uses: ./

The file and folder structure used for the test above is shown below.

.
├── .gitattributes
├── .github
│   └── workflows
│       └── main.yaml
├── .gitignore
├── .pre-commit-config.yaml
├── .taskcat.yml
├── Dockerfile
├── LICENSE
├── README.md
├── action.yaml
├── commitlint.config.js
└── templates
    └── my-template.yaml

[ShahradR]

We currently aren't able to provision an AWS account for end-to-end testing (see #3). Since the remaining work in this MVP depends on taskcat's ability to connect to an AWS account and create CloudFormation resources, this issue is on hold until #3 is resolved.

[ShahradR]

The AWS account creation issue outlined in #3 has been resolved by adding the user to the Service Catalog portfolio. Additionally, I wrote a quick gist around AWS Control Tower setup, for future reference.

[ShahradR]

Commit 27f8287 removes the inputs section of the action.yaml file, allowing the AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY environment variables to be properly passed to the taskcat Docker container.

GitHub Action inputs are prefixed with INPUT_. For example, when passing the AWS_ACCESS_KEY_ID, it would appear to the container as INPUT_AWS_ACCESS_KEY_ID, which taskcat does not recognize.

By removing the inputs from the file, the environment variables are passed as-is, resolving the issue.

In the .github/workflows/main.yaml file that consumes the action, we also need to change how the credentials are passed in. The changes are highlighted below.

Before change

---
on: [push]

jobs:
  taskcat:
    runs-on: ubuntu-latest
    name: taskcat
    steps:
      - name: taskcat
        id: taskcat
        uses: ./
        with:
          aws-access-key-id: ${{secrets.AWS_ACCESS_KEY_ID}}
          aws-secret-access-key: ${{secrets.AWS_SECRET_ACCESS_KEY}}

After change

---
on: [push]

jobs:
  taskcat:
    runs-on: ubuntu-latest
    name: taskcat
    steps:
      - name: Run taskcat
        id: taskcat
        uses: ./
        env:
          AWS_ACCESS_KEY_ID: ${{secrets.AWS_ACCESS_KEY_ID}}
          AWS_SECRET_ACCESS_KEY: ${{secrets.AWS_SECRET_ACCESS_KEY}}
          AWS_SESSION_TOKEN: ${{secrets.AWS_SESSION_TOKEN}}

[ShahradR]

Note that because I am using landing zones with the AWS Security Token Service (STS), I also need to pass the AWS_SESSION_TOKEN environment variable in addition to the AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY variables.

In my case, the session token can be retrieved by visiting the AWS Single Sign-On page, and clicking on Command line or programmatic access.

[ShahradR]

Opened #5 after determining that act is unable to render certain outputs. Tried running the action through GitHub itself to see if the output limitation is limited to act. However, the action.yaml file is incorrectly referencing the Docker container, preventing the workflow from even running. #6 was opened to track that work.

[ShahradR]

#6 has been resolved by properly referencing the Dockerfile in the action.yaml.

[ShahradR]

With the Dockerfile updated to pipe taskcat output to cat, act now works as expected. #5 has been resolved and closed.

[github-actions]

🎉 This issue has been resolved in version 1.0.0-develop.1 🎉

The release is available on GitHub release

Your semantic-release bot 📦🚀

[github-actions]

🎉 This issue has been resolved in version 1.0.0-develop.1 🎉

The release is available on GitHub release

Your semantic-release bot 📦🚀

[github-actions]

🎉 This issue has been resolved in version 1.0.0 🎉

The release is available on GitHub release

Your semantic-release bot 📦🚀