kubespray (ubuntu server)

1. Clone the kubespray repo

    git clone https://github.com/kubernetes-sigs/kubespray
    cd kubespray
   

2. (optional) Install pip if not yet installed

    sudo apt install python3-pip
   

3. Install required components

    sudo pip install -r requirements.txt
   

In case of error error: externally-managed-environment use pip's argument --break-system-packages:

    sudo pip install --break-system-packages -r requirements.txt

4. Copy example and create new inventory for Ansible

    cp -rfp inventory/sample inventory/mycluster
   

5. Declare your nodes IP addresses

    declare -a IPS=(192.168.10.81 192.168.10.82 192.168.10.83)
   

6. Prepare inventory

    CONFIG_FILE=inventory/mycluster/hosts.yaml python3 contrib/inventory_builder/inventory.py ${IPS[@]}
   

7. (optional) if an error No module named 'ruamel' occured, install missing package and try again to create the inventory.

    pip3 install ruamel.yaml
   

   or add argument --break-system-packages or you can use directly apt to install the package:

    sudo apt install python3-ruamel.yaml
   

8. Edit file inventory/mycluster/group_vars/k8s_cluster/k8s-cluster.yml where in line 70 (kube_network_plugin) change calico to cilium (here description about network addons compatibility and why we change Calico to another plugin)

9. Edit file inventory/mycluster/hosts.yaml where line 19 should be commented, because we do not want to have 2 master nodes.

    # node2
   

10. Run the Ansible playbook to create the cluster

     ansible-playbook -K -i inventory/mycluster/hosts.yaml --become --become-user=root cluster.yml
    

    Argument -K allows you to input the password

11. (optional) if something goes wrong you can reset your installation:

      ansible-playbook -K -i inventory/mycluster/hosts.yaml --become --become-user=root reset.yml
    

12. Copy kube config to local machine. File location is <MasterNodeIpAddress>:/root/.kube/config

13. Edit this file and in line 5 replace IP address with IP of your master node (node1).

14. Copy kube config to default location ~/.kube/config (folder may not yet exist) or set env variable KUBECONFIG with file location:

     export KUBECONFIG=~/config
    

15. (optional) install kubectl

https://kubernetes.io/docs/tasks/tools/install-kubectl-linux/

    curl -LO "https://dl.k8s.io/release/$(curl -L -s https://dl.k8s.io/release/stable.txt)/bin/linux/amd64/kubectl"
    sudo install -o root -g root -m 0755 kubectl /usr/local/bin/kubectl
    kubectl version --client

15. Now you should be able to run command on your k8s cluster. For example:

     kubectl get node
    

k8s examples

kubectl run hostinfo --image=finconreply/hostinfo
kubectl expose pod hostinfo --port 8080 --target-port 8080

kubectl create deployment hostinfo --image=finconreply/hostinfo --replicas 3
kubectl expose deployment hostinfo --port 8080 --target-port 8080

kubectl port-forward svc/hostinfo 38000:8080

kubectl delete service hostinfo
kubectl delete deployment hostinfo

kubectl create -f deployment.yaml
kubectl create -f service.yaml

kubectl port-forward svc/hostinfo 38000:8080

kubectl delete -f service.yaml
kubectl delete -f deployment.yaml