K8S CLUSTER

Kubernetes Multi Master Cluster

• kubectl/v1.25.3
• calico/cni:v3.24.5
• kubernetesui/dashboard:v2.7.0
• kubernetesui/metrics-scraper:v1.0.8
• open-iscsi
• helm

PREREQUISITES

You need to have your ssh id already provisioned; otherwise run: ssh-keygen -t rsa

Make sure you have copied your ssh key from ansible to the hosts prior to running playbooks, in our case - the vm template built in playbook 0- had both ssh installed and the key in authorized_keys

Or run the ansible-playbook command with -k to be prompted for password

DOCS

k8s-ha:

• 10.20.40.20 LOAD BALANCED (HA PROXY) VIRTUAL IP (KEEPALIVED)
• 10.20.40.21 k8s-ha1 12vCPU | 16GB RAM | 24GB + 100GB DISK (Cluster Master + Worker)
• 10.20.40.22 k8s-ha2 12vCPU | 16GB RAM | 24GB + 100GB DISK (Cluster Master + Worker)
• 10.20.40.23 k8s-ha3 12vCPU | 16GB RAM | 24GB + 100GB DISK (Cluster Master + Worker)
• 10.20.40.30 k8s-halb1 4vCPU | 4GB RAM | 24GB (External Round Robbin Load Balancer)
• 10.20.40.29 k8s-halb2 4vCPU | 4GB RAM | 24GB

INSTALL

Edit to your needs each file found:

grep 10.20.40 * -R | awk '{print $1}' | sort |uniq

Make sure you change everything exactly. Best to use find / replace

Note:

This playbook is designed to run with a template of Debian 11 fresh from netinst, which is automatically provisioned in playbook 0_create_vms.yaml from a ready made Appliance NOT provided. If you want to ansible to provision your VMs a template must be created in XenCenter/XCP-ng Center call it "Debian 11".

Therefore, if you don't have XenServer / XCP-ng edit start-k8s-ha-cluster.yaml and add "when: false" to first task called "Create vms using xe (XCP-ng / XenServer"

AND

Create your VM's with the default values k8s-ha or with your own edited version

Then once your vms are up contactable and ssh-able as root from your ansible host

From inside ansible directory run:

ansible-playbook -i hosts start-k8s-ha-cluster.yaml

KUBECTL ACCESS

ssh to 10.20.40.21-23 to run kubectl or helm - (helm repo do not sync between nodes)

• If you have kubectl locally scp 10.20.40.21:/etc/kubernetes/admin.conf ~/.kube/config

CONFIRM PODS

kubectl get pods -A

TO VIEW THE DASHBOARD

You need the NodePort from the Kubernetes Dashboard service, to find it:

kubectl describe service -nkubernetes-dashboard | grep -i nodeport |grep unset

Open up your browser to any of the nodes

https://10.20.40.21:< NodePort >/

https://10.20.40.22:< NodePort >/

https://10.20.40.23:< NodePort >/