Deploy Helm Apps

[marcellodesales]

Requirement

• Given a the deployment model, specify a Helm Application to be deployed.
• "Any" application from https://hub.kubeapps.com/ should be deployed automatically.

Design

• Specify the name of the Helm Chart to deploy
• Identify the type of ingress instance the cluster contains
  • Use default ingress nginx
  • This provides a default ingress to the cloud provider
• Identify if the cluster has tls secret
  • This provides HTTPS access to the service
• Identify if the cluster has external-dns
  • This provides Route53 endpoint to the service

[0x777]

Currently, gitkube simplifies this workflow

docker build -t ...
docker push ...
kubectl set-image ... 

with

git push remote master

What is your current workflow with a kubeapps application and how do you like to see it automated?

[marcellodesales]

@0x777 So, kubeapps offers a way to deploy a Helm Application from a given Helm Chart Registry. Given that and your workflow, what I was thinking is that we could do the following:

git push remote branch

Still should guarantee what gitkube does already, but for Helm Apps, as follows:

1. Identify if there's a Helm Spec
2. Compute the actual URL of the application based on the Cluster Config + External DNS name
   • Through external dns, we get the endpoint for the cluster Cluster DNS cluster.name.com
   • With the helm settings ingress.host, we can compute the URL myapp.cluster.name.com
3. Push the committed application to the associated Helm Chart Registry under the SHA version associated with the commit
   • Pushes to the registry with a known URL for installation
   • Add extra metadata for the branch, tag, etc
   • Final url for the app could be either myapp.cluster.name.com, for master branch, or SHA.myapp.cluster.name.com for the SHA, develop.myapp.cluster.name.com for the develop branch, etc.
   • The mapping for a gitflow mechanism should be provided in a metadata file like gitkube.yml
4. Execute the helm install with the URL for installation above