Getting Started with k0s and Terraform

This example shows how to create a k0s cluster in AWS using Terraform and then install Boundless Operator on it.

Prerequisites

Along with boundless CLI, you will also need the following tools installed:

• k0sctl - required for installing a k0s distribution
• terraform - for creating VMs in AWS

You will also need an AWS account and the AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, and AWS_SESSION_TOKEN env variables set for the AWS CLI.

Create virtual machines on AWS

Creating virtual machines on AWS can be easily done using the example Terraform scripts.

After copying the example TF scripts to your local machine, you can create the VMs with the following steps:

1. Create a terraform.tfvars file with content similar to:

cluster_name = "k0s-cluster"
controller_count = 1
worker_count = 1
cluster_flavor = "m5.large"
region = "us-east-1"

2. terraform init
3. terraform apply -auto-approve
4. terraform output --raw k0s_cluster > VMs.yaml

To get detailed information about the created VMs, use the AWS CLI:

aws ec2 describe-instances --region $(grep "region" terraform.tfvars | awk -F' *= *' '{print $2}' | tr -d '"')

Alternatively, for a visual overview: Go to the AWS EC2 page. Select the desired region from the dropdown menu at the top-right corner.

Install Boundless Operator on k0s

1. Download the example blueprint for creating a k0s cluster in AWS with TF

2. Edit the k0s-in-aws-with-tf.yaml blueprint to set the spec.kubernetes.infra.hosts values to those from the VMs.yaml file.

The spec.kubernetes.infra.hosts section should look similar to:

spec:
   kubernetes:
      provider: k0s
      version: 1.27.4+k0s.0
      infra:
      hosts:
      - ssh:
            address: 52.91.89.114
            keyPath: <TF examples folder>/aws_private.pem
            port: 22
            user: ubuntu
         role: controller
      - ssh:
            address: 10.0.0.2
            keyPath: <TF examples folder>/aws_private.pem
            port: 22
            user: ubuntu
         role: worker

3. Create the cluster:

bctl apply -f k0s-in-aws-with-tf.yaml

Note: bctl apply adds kube config context to default location and sets it as the current context

4. Update the cluster by modifying k0s-in-aws-with-tf.yaml and then running:

bctl update -f k0s-in-aws-with-tf.yaml

5. Monitor the status of the cluster’s Kubernetes pods with:

watch -n 1 kubectl get pods --all-namespaces

It will take a few moments before the pods are ready:

NAMESPACE          NAME                                                     READY   STATUS              RESTARTS   AGE
boundless-system   boundless-operator-controller-manager-677b86bdc4-rtjwb   1/2     Running             0          25s
boundless-system   helm-controller-79cc59c76b-vsr2v                         1/1     Running             0          5s
default            helm-install-nginx-mj2qt                                 0/1     ContainerCreating   0          3s
kube-system        coredns-878bb57ff-d4j99                                  1/1     Running             0          40s
kube-system        konnectivity-agent-jkz62                                 1/1     Running             0          39s
kube-system        kube-proxy-22rxj                                         1/1     Running             0          39s
kube-system        kube-router-mrbks                                        1/1     Running             0          39s
kube-system        metrics-server-7f86dff975-gs26h                          0/1     Running             0          40s

Accessing the cluster

The example app addon can now be accessed through the http://<controller-node-ip>:6443 URL.

Cleanup

Delete the cluster:

bctl reset -f k0s-in-aws-with-tf.yaml

Delete virtual machines by changing to the example TF folder and running:

terraform destroy --auto-approve