K8S Cluster Deployment via Bash

This project is under development as part of my personal Bash and Kubernetes studies. Temporarily does not work due to redesign; I am up to support multiple controllers + metalb

Current Stage

The current stage allows deploying a Kubernetes (K8s) cluster with multiple worker nodes and more than one controller. However, after deployment, any additional controller should be joined to the master via kubeadm, which is not automatically installed with this script.

Prerequisites

To run this script, you need several (at least 3) Debian-based hosts (either VMs or physical machines):

• 1 or more for a controller
• 1 or more for a worker
• 1 as a jumppad (a host used to generate initial K8s configs, certificates, etc., for further distribution across the hosts)

Requirements for all hosts:

• The same user credentials with superuser rights on all machines
• All hosts must have a network connection between each other
• You need to know the hostname for each host

Steps to Deploy

1. Clone the repository to the jumppad host:

   git clone https://github.com/RComrade/k8s_cluster_via_bash.git

2. Navigate to the project folder:

   cd k8s_cluster_via_bash

3. Run the deployment script:

   bash deploy.sh

What Happens Next?

• The script will automatically check for the latest K8s version
• The rest of the software versions are hardcoded at the beginning of the script
• You'll be prompted to confirm the necessary versions
• After the checks are passed, you'll be asked for machines.txt — an inventory file

Here's an example of its format:

```
controller-0 192.168.1.10 server-0.kubernetes.local server-0
controller-1 192.168.1.3 server.kubernetes.local server
worker-0 192.168.1.11 node-0.kubernetes.local node-0 10.200.0.0/24
worker-1 192.168.1.13 node-1.kubernetes.local node-1 10.200.1.0/24
worker-2 192.168.1.8 metalb.kubernetes.local metalb 10.200.2.0/24
```

The columns are:

• Role: Used for internal script logic; controller-0 is the master controller node, the rest of them are additional and should be added manually w/ kubeadm
• IPv4: The real IPv4 address of a host to be used in the cluster
• FQDN: The fully qualified domain name of a host (e.g., server-0.kubernetes.local), it can be customized but has been tested with kubernetes.local domain only; recommended to leave by default
• Hostname: The actual hostname of a host, set during OS setup or initial configuration
• Network (for workers only): This is required for starting containers within the worker; virtual network information (e.g., 10.200.0.0/24); recommended to leave by default

You will be asked to input the machines.txt details or provide the file manually in the project folder.

Summary

Once you've answered the initial prompts, the script should handle the rest of the setup

Feel free to reach out for feedback, questions, or bug reports

Telegram: @nvyatkin9154