CustomResourceDefinition for Helm

This is an experimental CRD controller for Helm releases.

You can use this to install, upgrade and delete charts in your cluster via regular Kubernetes API objects that look like:

apiVersion: helm.bitnami.com/v1
kind: HelmRelease
metadata:
  name: mydb
spec:
  # 'stable' repo
  repoUrl: https://kubernetes-charts.storage.googleapis.com
  chartName: mariadb
  version: 2.0.1
  values: |
    mariadbDatabase: mydb
    mariadbPassword: sekret
    mariadbRootPassword: supersekret
    mariadbUser: myuser

Advantages:

• Familiar. Integrates well with other tools like kubectl apply, kubecfg and declarative/gitops workflows.

• More secure. HelmRelease objects can be restricted via RBAC policy, including limiting access by namespace.

Install:

d=$HOME/.kube/plugins/helm
mkdir -p $d; cd $d
wget \
 https://raw.githubusercontent.com/bitnami-labs/helm-crd/master/plugin/helm/helm \
 https://raw.githubusercontent.com/bitnami-labs/helm-crd/master/plugin/helm/plugin.yaml
chmod +x helm

You now have a new kubectl plugin! See kubectl plugin helm --help for the new subcommands.

Run kubectl plugin helm init to perform the server-side install.

Server-side only

If you don't want (or need) the kubectl plugin, you can install the server-side components directly with:

kubectl apply -n kube-system -f deploy/tiller-crd.yaml

This will create the CRD, and replace(!) any existing kube-system/tiller-deploy with an unmodified tiller v2.9.1 release with the tiller port restricted and a new controller sidecar.

To use, start creating API objects similar to the example above.

FAQ

Does this replace helm CLI tool?

Only for the most basic operations. If you just want to consume (install/remove/update) public upstream charts, then yes.

If you want to do anything else then no, you will still need helm. In particular, installing a chart from local disk, or developing a chart requires helm and is unchanged with this controller.

Does this affect upstream chart development?

No. Charts themselves are unchanged, and chart development workflow and tools remains the same as before. In particular, developing charts still requires use of the helm CLI tool and "port forward" access to the tiller container.

Note that this last point implies full access over the cluster used for development (just as before this controller) and chart development is best performed against dedicated and disposable test clusters (minikube, etc).

Does this prevent helm CLI usage?

No. The helm CLI tool accesses tiller using a Kubernetes port-forward into the tiller pod, and this is unaffected by the presence of this controller. Port forward access can be blocked via RBAC policy (pods/portforward resource) if desired - and leaves HelmRelease resource objects as the only way to access tiller functionality.