This is one of my exploratory project to dive into DevOps practices. This is not a working "out of the box" example but more of a general reference repository

• Tested with nodejs v14.21.3 on Mac (intel)
• A Kubernetes (k8s) cluster (validated on remote, but should also be operable locally)
• ArgoCD installed in the cluster
• A container image repository
• A private Bitbucket repository to host Helm charts for all applications (lets call it helm charts repository)
• This Bitbucket repository (application repository) containing the TypeScript code, along with the application Helm chart, which gets copied over to the central Helm charts repository

• This repository represents 'awesome' application which runs as container / service in kubernetes. Each application will have its own repository and similar pipeline. helm charts repository will contain helm charts for each service

• The clone uses authentication token because it clones and pushes a private repository

npm install && npm run build

npm run test

When you update your code and push it to the application repository, the CI/CD pipeline does the following:

• If a tag is included (it should contain the application version) using git push --follow-tags:
  • The semver is validated
  • The container image is built and pushed
  • The Helm chart is updated with the new version and copied to the helm charts repository
• If a tag is not included, it checks if the Helm chart is updated. If it is, it gets copied to the helm charts repository
• ArgoCD picks up the Helm changes from the helm charts repository and deploys it to the cluster

Here are some additional functionalities and improvements you could consider:

• Have pipeline scripts in a seperate repository that uploads them to S3/GCS and pipeline downloads them and executes them. This will allow for central management of pipeline scripts (this is parcial solution that makes increasing number of application repositories maintanace a bit easier).
• Different flows for different tags
• Adding an ingress to the Helm chart (you should have an ingress controller)
• Reducing the size of the container
• Scanning code for security issues / lint
• Adding more services
• Incorporating and running tests in the CI/CD pipeline

Think something could be improved or have questions?
Found an error or oversight?
Feel free to open an issue or send a pull request.