Building Cluster API Provider for CloudStack
Prerequisites:
-
Follow the instructions here to install the following tools:
- docker
- kind
- kubectl
- clusterctl (Requires v1.1.5 +)
-
Create a local docker registry to save your docker image - otherwise, you need an image registry to push it somewhere else.
-
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
-
Set up Apache CloudStack credentials
-
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>
-
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 generatedinfrastructure-components.yaml
, where it gets passed to CAPC:export CLOUDSTACK_B64ENCODED_SECRET=$(base64 -w0 -i cloud-config)
-
-
Set the IMG environment variable so that the Makefile knows where to push docker image (if building your own)
export IMG=localhost:5000/cluster-api-provider-capc
make docker-build
make docker-push
-
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
-
Generate the CAPC manifests (if building your own) into
$RELEASE_DIR
make build
will generate and copyinfrastructure-components.yaml
and metadata.yaml files to$RELEASE_DIR
, which is./out
by default. You may want to override the default value withexport RELEASE_DIR=${HOME}/.cluster-api/overrides/infrastructure-cloudstack/<VERSION>/
to deploy the generated manifests for use by clusterctl before runningmake build
. -
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
-
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:
-
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
-
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
-
Apply the CAPC cluster spec to your kind management cluster
kubectl apply -f capc-cluster-spec.yaml
-
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
-
Get the generated kubeconfig for your newly created Apache CloudStack cluster
capc-cluster
clusterctl get kubeconfig capc-cluster > capc-cluster.kubeconfig
-
Install calico on the workload cluster so that pods can see each other
KUBECONFIG=capc-cluster.kubeconfig kubectl apply -f https://projectcalico.docs.tigera.io/manifests/calico.yaml
-
Verify the K8s cluster is fully up. (It may take a minute for the nodes status to all reach ready state.)
- 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
- Run
Validating the CAPC Cluster:
Run a simple kubernetes app called ‘test-thing’
- 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
- 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
% 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 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