Building Cluster API Provider for CloudStack

Prerequisites:

  1. Follow the instructions here to install the following tools:

    1. docker
    2. kind
    3. kubectl
    4. clusterctl (Requires v1.1.5 +)
  2. Create a local docker registry to save your docker image - otherwise, you need an image registry to push it somewhere else.

  3. Download this script into your local and run it. This script will create a kind cluster and configure it to use local docker registry:

    wget https://raw.githubusercontent.com/kubernetes-sigs/cluster-api/main/hack/kind-install-for-capd.sh
    chmod +x ./kind-install-for-capd.sh
    ./kind-install-for-capd.sh
    
  4. Set up Apache CloudStack credentials

    1. Create a file named cloud-config in the repo’s root directory, substituting in your own environment’s values

      [Global]
      api-url = <cloudstackApiUrl>
      api-key = <cloudstackApiKey>
      secret-key = <cloudstackSecretKey>
      
    2. Run the following command to save the above Apache CloudStack connection info into an environment variable, to be used by ./config/default/credentials.yaml and ultimately the generated infrastructure-components.yaml, where it gets passed to CAPC:

      export CLOUDSTACK_B64ENCODED_SECRET=$(base64 -w0 -i cloud-config)
      
  5. Set the IMG environment variable so that the Makefile knows where to push docker image (if building your own)

    1. export IMG=localhost:5000/cluster-api-provider-capc
    2. make docker-build
    3. make docker-push
  6. Set the source image location so that the CAPC deployment manifest files have the right image path in them in config/default/manager_image_patch.yaml

  7. Generate the CAPC manifests (if building your own) into $RELEASE_DIR

    make build will generate and copy infrastructure-components.yaml and metadata.yaml files to $RELEASE_DIR, which is ./out by default. You may want to override the default value with export RELEASE_DIR=${HOME}/.cluster-api/overrides/infrastructure-cloudstack/<VERSION>/ to deploy the generated manifests for use by clusterctl before running make build.

  8. Generate clusterctl config file so that clusterctl knows how to provision the Apache CloudStack cluster, referencing whatever you set for $RELEASE_DIR from above for the url:

    cat << EOF > ~/.cluster-api/cloudstack.yaml
    providers:
    - name: "cloudstack"
      type: "InfrastructureProvider"
      url: ${HOME}/.cluster-api/overrides/infrastructure-cloudstack/<VERSION>/infrastructure-components.yaml
    EOF
    
  9. Assure that the required Apache CloudStack resources have been created: zone, pod cluster, and k8s-compatible template, compute offerings defined (2GB+ of RAM for control plane offering with 2vCPU).

Deploying Custom Builds

Initialize the management cluster

Run the following command to turn your cluster into a management cluster and load the Apache CloudStack components into it.

clusterctl init --infrastructure cloudstack --config ~/.cluster-api/cloudstack.yaml

Creating a CAPC Cluster:

  1. Set up the environment variables. It will be populated by the values set here. See the example values below (and replace with your own!)

    The entire list of configuration variables as well as how to fetch them can be found here

    # The Apache CloudStack zone in which the cluster is to be deployed
    export CLOUDSTACK_ZONE_NAME=zone1
    
    # If the referenced network doesn't exist, a new isolated network
    # will be created.
    export CLOUDSTACK_NETWORK_NAME=GuestNet1
    
    # The IP you put here must be available as an unused public IP on the network
    # referenced above. If it's not available, the control plane will fail to create.
    # You can see the list of available IP's when you try allocating a public
    # IP in the network at
    # Network -> Guest Networks -> <Network Name> -> IP Addresses
    export CLUSTER_ENDPOINT_IP=192.168.1.161
    
    # This is the standard port that the Control Plane process runs on
    export CLUSTER_ENDPOINT_PORT=6443
    
    # Machine offerings must be pre-created. Control plane offering
    # must have have >2GB RAM available
    export CLOUDSTACK_CONTROL_PLANE_MACHINE_OFFERING="Large Instance"
    export CLOUDSTACK_WORKER_MACHINE_OFFERING="Small Instance"
    
    # Referring to a prerequisite capi-compatible image you've loaded into Apache CloudStack
    export CLOUDSTACK_TEMPLATE_NAME=kube-v1.23.3/ubuntu-2004
    
    # The SSH KeyPair to log into the VM (Optional: you must use clusterctl --flavor *managed-ssh*)
    export CLOUDSTACK_SSH_KEY_NAME=CAPCKeyPair6
    
    # Sync resources created by CAPC in Apache Cloudstack CKS. Default is false.
    # Requires setting CAPC_CLOUDSTACKMACHINE_CKS_SYNC=true before initialising the cloudstack provider.
    # Or set enable-cloudstack-cks-sync to true in the deployment for capc-controller.
    export CLOUDSTACK_SYNC_WITH_ACS=true
    
  2. Generate the CAPC cluster spec yaml file

    clusterctl generate cluster capc-cluster \
        --kubernetes-version v1.23.3 \
        --control-plane-machine-count=1 \
        --worker-machine-count=1 \
        > capc-cluster-spec.yaml
    
    
  3. Apply the CAPC cluster spec to your kind management cluster

    kubectl apply -f capc-cluster-spec.yaml
    
  4. Check the progress of capc-cluster, and wait for all the components (with the exception of MachineDeployment/capc-cluster-md-0) to be ready. (MachineDeployment/capc-cluster-md-0 will not show ready until the CNI is installed.)

    clusterctl describe cluster capc-cluster
    
  5. Get the generated kubeconfig for your newly created Apache CloudStack cluster capc-cluster

    clusterctl get kubeconfig capc-cluster > capc-cluster.kubeconfig
    
  6. Install calico or weave net cni plugin on the workload cluster so that pods can see each other

    KUBECONFIG=capc-cluster.kubeconfig kubectl apply -f https://raw.githubusercontent.com/projectcalico/calico/master/manifests/calico.yaml
    
    

    or

    KUBECONFIG=capc-cluster.kubeconfig kubectl apply -f https://raw.githubusercontent.com/weaveworks/weave/master/prog/weave-kube/weave-daemonset-k8s-1.11.yaml
    
    
  7. Verify the K8s cluster is fully up. (It may take a minute for the nodes status to all reach ready state.)

    1. Run KUBECONFIG=capc-cluster.kubeconfig kubectl get nodes, and observe the following output
    NAME                               STATUS   ROLES                  AGE     VERSION
    capc-cluster-control-plane-xsnxt   Ready    control-plane,master   2m56s   v1.20.10
    capc-cluster-md-0-9fr9d            Ready    <none>                 112s    v1.20.10
    

Validating the CAPC Cluster:

Run a simple kubernetes app called ‘test-thing’

  1. Create the container
KUBECONFIG=capc-cluster.kubeconfig kubectl run test-thing --image=rockylinux/rockylinux:8 --restart=Never -- /bin/bash -c 'echo Hello, World!'
KUBECONFIG=capc-cluster.kubeconfig kubectl get pods
  1. Wait for the container to complete, and check the logs for ‘Hello, World!’
KUBECONFIG=capc-cluster.kubeconfig kubectl logs test-thing

kubectl/clusterctl Reference:

  • Pods in capc-cluster -- cluster running in Apache CloudStack with calico cni
% KUBECONFIG=capc-cluster.kubeconfig kubectl get pods -A
NAMESPACE     NAME                                                       READY   STATUS      RESTARTS   AGE
default       test-thing                                                 0/1     Completed   0          2m43s
kube-system   calico-kube-controllers-784dcb7597-dw42t                   1/1     Running     0          4m31s
kube-system   calico-node-mmp2x                                          1/1     Running     0          4m31s
kube-system   calico-node-vz99f                                          1/1     Running     0          4m31s
kube-system   coredns-74ff55c5b-n6zp7                                    1/1     Running     0          9m18s
kube-system   coredns-74ff55c5b-r8gvj                                    1/1     Running     0          9m18s
kube-system   etcd-capc-cluster-control-plane-tknwx                      1/1     Running     0          9m21s
kube-system   kube-apiserver-capc-cluster-control-plane-tknwx            1/1     Running     0          9m21s
kube-system   kube-controller-manager-capc-cluster-control-plane-tknwx   1/1     Running     0          9m21s
kube-system   kube-proxy-6g9zb                                           1/1     Running     0          9m3s
kube-system   kube-proxy-7gjbv                                           1/1     Running     0          9m18s
kube-system   kube-scheduler-capc-cluster-control-plane-tknwx            1/1     Running     0          9m21s
  • Pods in capc-cluster -- cluster running in Apache CloudStack with weave net cni
%KUBECONFIG=capc-cluster.kubeconfig kubectl get pods -A
NAMESPACE     NAME                                                       READY   STATUS      RESTARTS       AGE
default       test-thing                                                 0/1     Completed   0              38s
kube-system   coredns-5d78c9869d-9xq2s                                   1/1     Running     0              21h
kube-system   coredns-5d78c9869d-gphs2                                   1/1     Running     0              21h
kube-system   etcd-capc-cluster-control-plane-49khm                      1/1     Running     0              21h
kube-system   kube-apiserver-capc-cluster-control-plane-49khm            1/1     Running     0              21h
kube-system   kube-controller-manager-capc-cluster-control-plane-49khm   1/1     Running     0              21h
kube-system   kube-proxy-8lfnm                                           1/1     Running     0              21h
kube-system   kube-proxy-brj78                                           1/1     Running     0              21h
kube-system   kube-scheduler-capc-cluster-control-plane-49khm            1/1     Running     0              21h
kube-system   weave-net-rqckr                                            2/2     Running     1 (3h8m ago)   3h8m
kube-system   weave-net-rzms4                                            2/2     Running     1 (3h8m ago)   3h8m
  • Pods in original kind cluster (also called bootstrap cluster, management cluster)
% kubectl  get pods -A
NAMESPACE                           NAME                                                             READY   STATUS    RESTARTS   AGE
capc-system                         capc-controller-manager-55798f8594-lp2xs                         1/1     Running   0          30m
capi-kubeadm-bootstrap-system       capi-kubeadm-bootstrap-controller-manager-7857cd7bb8-rldnw       1/1     Running   0          30m
capi-kubeadm-control-plane-system   capi-kubeadm-control-plane-controller-manager-6cc4b4d964-tz5zq   1/1     Running   0          30m
capi-system                         capi-controller-manager-7cfcfdf99b-79lr9                         1/1     Running   0          30m
cert-manager                        cert-manager-848f547974-dl7hc                                    1/1     Running   0          31m
cert-manager                        cert-manager-cainjector-54f4cc6b5-gfgsw                          1/1     Running   0          31m
cert-manager                        cert-manager-webhook-7c9588c76-5m2b2                             1/1     Running   0          31m
kube-system                         coredns-558bd4d5db-22zql                                         1/1     Running   0          48m
kube-system                         coredns-558bd4d5db-7g7kh                                         1/1     Running   0          48m
kube-system                         etcd-capi-test-control-plane                                     1/1     Running   0          48m
kube-system                         kindnet-7p2dq                                                    1/1     Running   0          48m
kube-system                         kube-apiserver-capi-test-control-plane                           1/1     Running   0          48m
kube-system                         kube-controller-manager-capi-test-control-plane                  1/1     Running   0          48m
kube-system                         kube-proxy-cwrhv                                                 1/1     Running   0          48m
kube-system                         kube-scheduler-capi-test-control-plane                           1/1     Running   0          48m
local-path-storage                  local-path-provisioner-547f784dff-f2g7r                          1/1     Running   0          48m