Skip to content

Latest commit

 

History

History
37 lines (31 loc) · 2.04 KB

README.md

File metadata and controls

37 lines (31 loc) · 2.04 KB
Raw

Pipelines: AWS

This project intended to implement a simple Continuous Delivery Pipeline, based on AWS CodePipeline.

The repository is configured with GitHub Pages. The deployment should have consisted of putting the content of the main branch into gh-pages branch which publishes it to the website.

The BASH snippets used in various tasks are implemented in pipeline.sh script.

The pipeline was intended to have four stages

  • Spell check - checks the spelling of all markdown files in this repository
  • Link check - checks the links of all markdown files in this repository
  • Release creation - the pipeline, when run on main branch, creates a new Git tag and GitHub release. A new release is created only if one does not exist yet for the given commit
  • Release deployment - the pipeline, when run on main branch, deploys release by copying markdown files into gh-pages branch. The deployment happens only if the markdown files have been updated. The commit message in gh-pages branch contains the release version number

UNFORTUNATELY AWS CodePipeline and CodeBuild does not clone the Git repository it builds and so the .git metadata is not available. This means that the creation of Git tag or merging of Git branches is not possible.

To make it possible additional commands would have to be added to clone the repository. This in turn would require to hardcode in the buildspec.yml the Git repository URL as none is available in environment variables and the Git branch name (e.g. to differentiate a build of main branch from a build of pull request branch. Finally, to push Git commits credentials would also have to be provided and handled. All of this, while technically possible, requires quite a bit of hardcoding and scripting. Having to do this means that AWS CodePipeline and CodeBuild are not suited for scenarios where CI/CD job needs to perform Git operations.