kubernetes-playbooks

Ansible playbooks that creates a Kubernetes 1.23 cluster of Openstack instances running Ubuntu 20.04 LTS.

Prerequisites

Ansible and Python3 installed on the local machine (# yum install ansible).
An OpenStack security group for SSH and ICMP access named SSH and ICMP.
Terraform and OpenStack CLI tools installed on the local machine.

Create a `keystone_rc` file

$ cp keystone_rc.sh.example keystone_rc.sh

$ chmod 0600 keystone_rc.sh

The keystone_rc.sh file will contain your API password so be careful with where you store it, and make sure it's private. Once it is, add your API password for OpenStack. You can also modify the worker count, network version, etc.

Then load the file to the shell environment on the local computer:

$ source keystone_rc.sh

Add a public key for the cluster to the API user

SSH key pairs are tied to users, and the dashboard and API user are technically different. The SSH public key therefore has to be added to the API user explicitly:

$ openstack keypair create --public-key /path/to/keyfile.pub k8s-nodes

Build the cluster using Terraform

Change directory to tf-project and initialize Terraform:

$ cd tf-project

$ terraform init

Then verify, plan and apply with Terraform:

$ terraform validate

$ terraform plan

$ terraform apply

Change directory back to the main directory:

$ cd ..

Create an inventory file for Ansible

After creating the cluster on OpenStack, Terraform created a ansible_inventory file in the tf-project directory. It contains the machine names and IP addresses for the cluster.

Alternatively, a hosts file can be created. Add the IP address to the master and workers in the hosts file using a text editor, and make sure each machine can be reached using SSH:

$ cp hosts.example hosts

$ vim hosts

Install Kubernetes dependencies on all servers

$ ansible-playbook -i tf-project/ansible_inventory playbooks/kube-dependencies.yml

Initialize the master node

$ ansible-playbook -i tf-project/ansible_inventory playbooks/master.yml

ssh onto the master and verify that the master node get status Ready:

$ ssh -i /path/to/ssh-key ubuntu@<master_ip>
ubuntu@k8s-master-1:~$ kubectl get nodes
NAME           STATUS   ROLES                  AGE   VERSION
k8s-master-1   Ready    control-plane,master   59s   v1.23.4

Add the worker nodes

$ ansible-playbook -i tf-project/ansible_inventory playbooks/workers.yml

Run kubectl get nodes once more on the master node to verify the worker nodes got added.

Change or destroy the cluster

Edit cluster settings in the keystone_rc.sh and source it again before re-running terraform apply to change the cluster, before re-running the playbooks to add new workers.

Destroy the cluster when done:

$ cd tf-project

$ terraform destroy

Credits

Based on bsder's Digital Ocean tutorial «How To Create a Kubernetes 1.11 Cluster Using Kubeadm on Ubuntu 18.04».

License

See the LICENSE file for license rights and limitations (MIT).

asmtal / kubernetes-playbooks Goto Github PK