Advanced Options¶

This guide covers advanced configuration and customisation scenarios beyond the defaults.

Adjusting Node Resources¶

Increase Worker RAM¶

For resource-intensive workloads, increase worker memory in cluster.yaml:

- name: "node01"
  ip: "192.168.56.11"
  memory: 4096          # 4 GB instead of default 1.5 GB
  cpus: 2               # 2 CPUs instead of default 1
  role: "worker"

After changing resources, destroy and recreate the affected node:

vagrant destroy node01 -f
vagrant up node01

Warning

vagrant reload does not apply memory/CPU changes. You must destroy and recreate the VM.

Changing Network Configuration¶

Custom Pod CIDR¶

The default pod CIDR is 10.244.0.0/16. To use a different range:

cluster:
  pod_cidr: "172.16.0.0/16"

Full Cluster Rebuild Required

Changing the pod CIDR requires destroying all VMs and re-provisioning from scratch, since it is baked into kubeadm init and the Calico configuration.

Custom Service CIDR¶

cluster:
  service_cidr: "10.100.0.0/16"

The same full-rebuild requirement applies.

Scaling the Cluster¶

Adding Worker Nodes to a Running Cluster¶

Add a new node entry to cluster.yaml:

- name: "node04"
  ip: "192.168.56.14"
  memory: 2048
  cpus: 2
  role: "worker"

Bring up only the new node:
```
vagrant up node04
```
The new worker will automatically find and execute the join command from the shared /vagrant folder.

Join Token Expiry

kubeadm join tokens expire after 24 hours. If the cluster was created more than 24 hours ago, generate a new token first:

vagrant ssh controlplane
sudo kubeadm token create --print-join-command > /vagrant/join-command.sh

Removing a Worker Node¶

Drain the node from the control plane:

vagrant ssh controlplane
kubectl drain node03 --ignore-daemonsets --delete-emptydir-data
kubectl delete node node03

Destroy the VM:
```
vagrant destroy node03 -f
```
Remove or comment out the node entry in cluster.yaml.

Changing the Calico CNI Version¶

To use a different Calico version:

cluster:
  calico_version: "v3.28.0"

Note

This requires a full cluster rebuild. The Calico manifests are downloaded from the official release URL based on this version string.

Using a Different Vagrant Box¶

To use a different base OS image:

cluster:
  box: "ubuntu/focal64"    # Ubuntu 20.04 LTS

Compatibility

The provisioning scripts are tested with ubuntu/jammy64 (22.04 LTS). Other Ubuntu versions may work but are not officially tested. Non-Ubuntu distributions are not supported in v0.1.0.

Useful kubectl Commands¶

# Detailed node information
kubectl describe node controlplane

# Resource usage (requires metrics-server, not included in v0.1.0)
kubectl top nodes
kubectl top pods -A

# Watch events in real time
kubectl get events --watch

# Get cluster information
kubectl cluster-info

# View kubeadm cluster configuration
kubectl -n kube-system get cm kubeadm-config -o yaml