cluster-controller

README

cluster-controller

This repository implements a controller for watching Cluster resources as defined with a CustomResourceDefinition (CRD).

This is a WIP repo for a POC, TODO alter directory structure and create a CI job to handle codegen, see: https://blog.openshift.com/kubernetes-deep-dive-code-generation-customresources/

Purpose

This custom controller manages a custom resource of type KrakenCluster.

Deploy CRD and create a resource

If your cluster has RBAC enabled you will need to create a clusterrole and clusterbinding:

$ kubectl create -f assets/clusterrole.yaml
$ kubectl create -f assets/clusterrolebinding.yaml

$ helm install chartmuseum/cluster-controller --version 0.1.3

You may then create a sample resource by updating either assets/test-aws.yaml or assets/test-maas.yaml.

For an aws cluster make sure you update the following fields in assets/test-aws.yaml:

metadata
  name : # name of your resource, will show up with $ kubectl get kc ,  should be unique for namespace
spec:
  cloudProvider:
    credentials:
      username: # aws username or iam profile name to be used
      accesskey: # aws access key id
      password: # aws secret access key
    region: # optional , but ideally would be set. ex = "aws/us-west-2"
  provisioner:
    bundle: # juju bundle desired for kubernetes cluster creation. ex - cs:bundle/kubernetes-core-306
  cluster:
    clusterName: # name of your cluster

For a maas cluster make sure you update the following fields in assets/test-maas.yaml:

metadata
  name : # name of your resource, will show up with $ kubectl get kc, should be unique for namespace
spec:
  cloudProvider:
    credentials:
      username: # maas username
      password: # maas api key
  provisioner:
    bundle: # juju bundle desired for kubernetes cluster creation. ex - cs:bundle/kubernetes-core-306
    maasEndpoint: # ex-"http://your-ip/MAAS/api/2.0"
  cluster:
    clusterName: # name of your cluster

Then run:

kubectl create -f assets/test-aws.yaml

or

kubectl create -f assets/test-maas.yaml

Monitor the KrakenCluster resources with:

kubectl get kc
kubectl describe kc <metadata.name>

Running Locally Without a Container

# assumes you have a working kubeconfig, not required if operating in-cluster
$ go run *.go -kubeconfig=$HOME/.kube/config -logtostderr=true

Changing the Specification for KrakenCluster

vi pkg/apis/clustercontroller/v1alpha1/types.go
hack/update-codegen.sh

To Test Locally With Minikube and the Helm Chart

Start a minikube cluster - set your Kubernetes version to v1.10.0, a bug in v1.9.2 (default) will cause you some volume grief.

$ minikube start --kubernetes-version v1.10.0

Start tiller: $ helm init

Make sure you have the chartmuseum repo with $ helm repo list. If not: $ helm repo add chartmuseum https://charts.cnct.io.

Once chartmuseum is added run $ helm repo update for good measure. You can check you have the most current version of our chart cluster-controller by running $ helm search chartmuseum. The most current helm chart version is pinned in the .versionfile in this repo's home directory.

Install the clusterrole and clusterbindings for the controller with kubectl: $ kubectl create -f assets/clusterrole.yaml $ kubectl create -f assets/clusterrolebinding.yaml

Ok! Tiller pods are running, clusterrole and clusterbindings are created, Helm has the most current version of our chart - let's install the chart!

$ helm install chartmuseum/cluster-controller

If there is not installation error, let's see if everything is running.

Helm chart status: $ helm list Check the controller pod: $ kubectl get po You can check the controller logs with $ kubectl logs <controller pod> to make sure the controller is running and check your cluster status.

Let's make sure the CRD is installed: $ kubectl get crd

Nothing blowing up? Let's try to deploy a new KrakenCluster object! In ./assets you will find a test-aws.yaml. Put your AWS or Maas creds ( specified above ) in that file, rename the cluster something meaningful and create it: $ kubectl create -f assets/test-aws.yaml

You can see/describe your new resource with $ kubectl get kc and watch the logs from the controller pod to check the latest status. You can also monitor the cluster with the AWS UI.

To delete your cluster run:

$ kubectl delete kc <my-kc-resource>

NOTE: Delete your kc resources BEFORE you delete the cluster-controller. If you don't you will not be able to delete your kc resource or your KrakenCluster crd resource.