Скачать или смотреть This is how you run AI pods on GPU nodes - Kubernetes

In this video, we are going to challenge everything we have learned so far and more 😅. We will:

Add a new node to an existing kubernetes cluster
Set the new node as GPU node and configure nvidia plugin on it
Create a new helm chart for our new AI service
Configure our python container to use CUDA

This was fun, but long. If you want to try something similar and you got into issues, please let me know 😊

Timstamps:

00:00:00 Intro
00:01:46 Code overview
00:08:20 Preparation for adding the new node
00:13:14 Running the scale playbook
00:15:13 Giving the playbook different users for different machines
00:17:39 Kubespray config for Pop_OS!
00:25:10 Taints and Tolerations on GPU node
00:26:46 Add the nvidia plugin to kubernetes
00:34:19 Installing latest nvidia container toolkit for containerd
00:37:06 Creating a helm chart for Tiba
00:43:25 Add the new chart to our existing Flux
00:49:22 Containerizing the AI service
00:53:24 Mounting the voice models in the container
00:58:05 Get CUDA running inside the container
01:01:10 Outro


Want to see more contents like this? 📖 🤓
Follow me on:

Medium:   / fdn-sharp  
LinkedIn:   / farhad-nowzari-94060699  
Github: https://github.com/farhadnowzari

Have fun coding ❤️