Last modified May 25, 2023

Create a cluster using cluster API in AWS (CAPA)

In order to create a cluster in an installation using the new cluster API in AWS environment you can follow these simple steps:

Create yaml files

Create a template for a yaml file using kubectl gs template cluster command. An example of the command would be:

$ kubectl gs template cluster --provider capa --organization YOUR_ORG > cluster.yaml

You can check all the available parameters in the command reference

This command will return several objects that you need to create in your management cluster to create your cluster. You can now edit the file to include your preferred options. Some interesting things you can customize here are:
- instance type: default value m5.xlarge
- size of the node and master node pool
- availability zone usage

---
apiVersion: v1
data:
values: |
  aws: {}
  bastion: {}
  clusterName: brhf3
  controlPlane:
    replicas: 3
  machinePools:
    machine-pool0:
      availabilityZones:
      - eu-central-1a
      instanceType: m5.xlarge
      maxSize: 10
      minSize: 3
      rootVolumeSizeGB: 300
  network:
    availabilityZoneUsageLimit: 3
  organization: YOUR_ORG  
kind: ConfigMap
metadata:
creationTimestamp: null
labels:
  giantswarm.io/cluster: brhf3
name: brhf3-userconfig
namespace: org-YOUR_ORG
---
apiVersion: application.giantswarm.io/v1alpha1
kind: App
metadata:
labels:
  app-operator.giantswarm.io/version: 0.0.0
name: brhf3
namespace: org-YOUR_ORG
spec:
catalog: cluster
config:
  configMap:
    name: ""
    namespace: ""
  secret:
    name: ""
    namespace: ""
kubeConfig:
  context:
    name: ""
  inCluster: true
  secret:
    name: ""
    namespace: ""
name: cluster-aws
namespace: org-YOUR_ORG
userConfig:
  configMap:
    name: brhf3-userconfig
    namespace: org-YOUR_ORG
version: 0.9.2
---
apiVersion: v1
data:
values: |
  clusterName: brhf3
  organization: YOUR_ORG  
kind: ConfigMap
metadata:
creationTimestamp: null
labels:
  giantswarm.io/cluster: brhf3
name: brhf3-default-apps-userconfig
namespace: org-YOUR_ORG
---
apiVersion: application.giantswarm.io/v1alpha1
kind: App
metadata:
labels:
  app-operator.giantswarm.io/version: 0.0.0
name: brhf3-default-apps
namespace: org-YOUR_ORG
spec:
catalog: cluster
config:
  configMap:
    name: ""
    namespace: ""
  secret:
    name: ""
    namespace: ""
kubeConfig:
  context:
    name: ""
  inCluster: true
  secret:
    name: ""
    namespace: ""
name: default-apps-aws
namespace: org-YOUR_ORG
userConfig:
  configMap:
    name: brhf3-default-apps-userconfig
    namespace: org-YOUR_ORG
version: 0.5.4

Creating the cluster

Applying this files to the management cluster is enough to create the cluster. When this happens cluster operator takes control and creates the necessary resources for the new workload cluster.

  kubectl apply -f cluster.yaml

You can check that the cluster has been properly created using kubectl get clusters in the management cluster.

kubectl get clusters -n org-YOUR_ORG 
NAME    PHASE         AGE     VERSION
brhf3   Provisioned   10m

Need help, got feedback?

We listen to your Slack support channel. You can also reach us at support@giantswarm.io. And of course, we welcome your pull requests!