This project creates a kubernetes cluster on Raspberry Pi servers. I use if for learning and experimenting rather than hosted services as there is no magic happening under the hood that you can't access.
The setup uses keepalived and haproxy even in a single controller configuration. With multiple controllers,
the configurations of these services are modified to support multiple controllers.
The minimum cluster size is two nodes - one controller and one worker.
This is based on instructions from here:
https://kubernetes.io/docs/setup/production-environment/tools/kubeadm/create-cluster-kubeadm/
- 3x RaspiberryPi 4 8GByte
- Ubuntu Server 22.04.1 LTS (64 bit)
- Kubernetes v1.25.3
- Terraform v1.3.2
- Ansible [core 2.13.5]
Install the OS image using the Imager tool from here.
Once you have it downloaded:
- choose the OS as 'Ubuntu Server 22.04.1 LTS (64 bit)'
- set the username/password and ssh public key information in the settings
- write to the SD disk
Connect the servers to a wired network and start them all. Once they are all up and you can ssh into them, it's time for the next step.
Terraform is used to build the ansible configurations.
Copy the file terraform.tfvars.example to terraform.tfvars and configure to match your installation.
Then run:
terraform init
terraform apply
The ansible scripts are ready to run at this point.
./local/ansible/run_ansible.sh
Once this completes, you should have a working kubernetes cluster with the dashboard installed.
Note that it can take some time from when ansible finishes to when the cluster is ready.
Copy the kubectl config file from the crontroller at /root/.kube/config file to your local machine
at ~/.kube/config.
This allows you to run kubectl commands locally. Test with:
kubectl get nodes
kubectl get pods --all-namespaces
Documentation on the dashboard and setup can be found here.
Most of the setup is completed by ansible except for the few manual steps to follow below.
Create a token to login with. You will need this once you connect to the service.
kubectl -n kubernetes-dashboard create token admin-user
Start the proxy:
kubectl proxy
Browse to dashboard:
http://localhost:8001/api/v1/namespaces/kubernetes-dashboard/services/https:kubernetes-dashboard:/proxy/