Kubernetes lab for CKA study

This is a simple set of configurations, Ansible tasks & playbooks to deploy & manage virtual machines for the purposes of studying for the CKA exam.

Targeting the current CKA exam version of Kubernetes, v1.24.4.

Topology

32GB + Intel Xeon E1231v3 CPU on the host machine.

Six (6) virtual machines act as the cluster.

• Three (3) Ubuntu 22.04 VMs acting as cluster control plane nodes and stacked etcd. 3GB/2CPU
  • kubeadm, kubelet and kubectl
  • keepalived + haproxy for high availability. See options for software load balancing for details.
• Three (3) Ubuntu 22.04 VMs acting as worker nodes. 2GB/2CPU
  • kubelet

Each server has two NICs, one host-only private network (192.168.56.0/24) and the other bridged to the host for Internet access (192.168.1.0/24).

An additional VM is used as an Ansible control host and development.

Usage

It's assumed that you have a valid Ansible inventory specified in /etc/ansible/hosts with two groups, controlplane and nodes. There is also a provided hosts file that can be specified with the -i option.

[kubernetes]
ubuntu0[1:6].lab

[controlplane]
ubuntu0[1:3].lab

[nodes]
ubuntu0[4:6].lab

I'm also using the common role to configure the /etc/hosts file for internal name resolution.

Playbook follows the kubeadm installation instructions found here: https://kubernetes.io/docs/setup/production-environment/tools/kubeadm/install-kubeadm/

$ ansible-playbook -i inventory/hosts site.yaml

Upgrades

The following tags can be used to upgrade specific components and remove dpkg hold.

Note: This playbook does not drain any nodes of active pods.

To upgrade a cluster:

1. Change kubernetes_version in group_vars/all or set it via extra-args.

2. Run kubeadm & kubectl upgrades on a control plane node. Use --limit <nodename> to choose which one to perform the upgrade on.

   ansible-playbook -i inventory/hosts site.yaml --tags "upgrade,kubeadm,kubectl" --limit ubuntu01.lab
   ansible-playbook -i inventory/hosts site.yaml --tags "upgrade,kubeadm,kubectl" --limit ubuntu02.lab
   ansible-playbook -i inventory/hosts site.yaml --tags "upgrade,kubeadm,kubectl" --limit ubuntu03.lab

3. Upgrade the kubectl on the control plane nodes.

   ansible-playbook -i inventory/hosts site.yaml --tags "upgrade,kubelet" --limit "controlplane"

4. Drain the nodes.

   ansible nodes -i inventory/hosts --become -a 'kubectl drain $(hostname)'

5. Upgrade the nodes.

   ansible-playbook -i inventory/hosts site.yaml --tags "upgrade,kubeadm,kubectl,kubelet" --limit "nodes"

6. Uncordon the nodes.

   ansible nodes -i inventory/hosts --become -a 'kubectl uncordon $(hostname)'

Rest

Use kubeadm reset via ansible ad-hoc to completely reset a cluster node.

ansible kubernetes --become -i inventory/hosts -a 'kubeadm reset -f --ignore-preflight-errors all'