Build a High Availability Kubernetes Cluster on Low-Power Intel NUCs

Table of Contents

1. Introduction
2. Choosing the Hardware for the Kubernetes Cluster
3. Rack Mounting the Intel NUCs
4. Remote Controlling the Devices
5. Installing the Operating System
6. Setting up Kubernetes
7. Deploying Workloads and testing HA
8. Conclusion

Setting up a Highly Available Kubernetes Cluster on Low-Power Intel NUCs

In recent years, I have been running virtualized clusters in my home lab using Proxmox. However, I recently decided to switch things up and run my Kubernetes cluster on a set of low-power Intel NUCs that I had. These small yet powerful devices, with their compact form factor and impressive performance, made them the perfect choice for my project.

1. Introduction

In this article, I will share my experience and provide a step-by-step guide on setting up a highly available Kubernetes cluster using low-power Intel NUCs. From choosing the right hardware to configuring the cluster, I will cover all the necessary steps to get you up and running.

2. Choosing the Hardware for the Kubernetes Cluster

Before diving into the setup process, it is important to choose the right hardware for your Kubernetes cluster. In my case, I opted for Intel NUCs due to their small form factor and powerful processors. I selected NUCs with Intel Core i7 processors and maxed out the RAM on each device to ensure optimal performance.

3. Rack Mounting the Intel NUCs

To make the most of the space in my home lab, I decided to rack mount the Intel NUCs. I found a rack mount system designed specifically for small form factor devices like the NUCs. This kit provided a secure and organized solution, ensuring easy cable management and keeping the devices securely in place.

4. Remote Controlling the Devices

To remotely control the Intel NUCs, I used a Pi KVM along with an HDMI switch. This setup allowed me to connect multiple devices to the Pi KVM and power them on using Wake-on-LAN. Additionally, I had a Unifi Smart PDU for power control when Wake-on-LAN was not successful.

5. Installing the Operating System

For the operating system, I chose Ubuntu Server, as it aligned with the rest of my infrastructure. Using Metal as a Service, I was able to boot and provision the NUCs effortlessly. I created static IP addresses for each device and configured DNS entries for easy management.

6. Setting up Kubernetes

Setting up Kubernetes can be a complex task, but I chose K3s for its lightweight nature and active community support. Rather than spending hours manually installing Kubernetes, I used an Ansible playbook that automated the entire process. This allowed me to quickly set up a highly available Kubernetes cluster with built-in high availability for API and service load balancers.

7. Deploying Workloads and Testing HA

With the Kubernetes cluster up and running, it was time to deploy workloads and test the high availability. I started by creating a Kubernetes deployment for a simple web server using Nginx. By setting the replicas to three and introducing topology constraints, I ensured that the workload was spread across all three nodes for high availability. I then exposed the service using MetalLB, which provided a virtual load balancer for accessing the web page.

8. Conclusion

In conclusion, setting up a highly available Kubernetes cluster on low-power Intel NUCs is a challenging yet rewarding project. With the right hardware, proper rack mounting, remote control capabilities, and the right software configuration, you can create a robust and efficient Kubernetes environment. Whether you are looking to host websites, Game servers, home automation systems, or any other workload, a well-optimized Kubernetes cluster on low-power Intel NUCs can handle it all.

By following the steps outlined in this article, you will be well on your way to setting up your own Kubernetes cluster and unlocking the full potential of these small yet mighty devices.

Highlights:

• Setting up a highly available Kubernetes cluster on low-power Intel NUCs
• Choosing the right hardware for optimal performance
• Rack mounting the Intel NUCs for efficient space usage
• Remote controlling the devices using Pi KVM and Unifi Smart PDU
• Installing Ubuntu Server as the operating system
• Setting up Kubernetes with K3s for lightweight and high availability
• Deploying workloads and testing high availability
• Creating a robust and versatile Kubernetes environment

FAQs:

Q: Can I use any other small form factor devices instead of Intel NUCs? A: Yes, you can use other small form factor devices like Mac Minis or Lenovo ThinkStations. The rack mount kits mentioned in the article are designed to accommodate various devices.

Q: Is it necessary to rack mount the Intel NUCs? A: Rack mounting the NUCs is not mandatory, but it provides a more organized and space-efficient solution, especially if you have multiple devices.

Q: Can I use a different operating system instead of Ubuntu Server? A: While Ubuntu Server is a popular choice, you can use other Linux distributions based on your preferences and compatibility with your infrastructure.

Q: Can I deploy workloads that are not stateless on the Kubernetes cluster? A: It is generally recommended to deploy stateless workloads for high availability out of the box. However, if your workload requires storage or state, you can configure external databases or other solutions to ensure data persistence.

Resources:

• Pi KVM
• Metal as a Service
• K3s
• MetalLB