Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Latest commit

 

History

History
55 lines (28 loc) · 1.61 KB

File metadata and controls

55 lines (28 loc) · 1.61 KB

K8s / K3d / Docker Setup

Install and create the cluster using docker and k3d


Install and Setup Docker CE

Docker Setup

Install Utils (Kubectl, Helm)

Utils install steps

Performance Tune Host Machine

Use the following settings to allow for heavy workloads on you host machine

Performance Tune

Install K3D

Use the official installation script to install the latest version of k3d:

K3D Install

Enable GPU in Cluster

Radeon GPU (ROCm): GPU devices are passed through to K3d as volume mounts (/dev/kfd, /dev/dri) during cluster creation — no custom K3s image required. Install the AMD k8s device plugin to make GPUs allocatable.

NVIDIA GPU: GPU support with NVIDIA drivers is part of the custom k3s image used when creating the k3d cluster. See the GPU readme for information on running the custom k3s image and building new custom images.

GPU setup steps

Install a Cluster Manager (Portainer Recommended)

Rancher is a resource hog and crashes a lot in a home lab setup. I suggest portainer as a simpler and more efficent cluster manager.
Portainer install steps OR Rancher install steps

Install ArgoCD

ArgoCD install steps

Install Dapr

Dapr install steps

Install Various Stacks and Applications

Stack install steps

Complete Cluster Teardown & Cleanup

Uninstall / cleanup steps