OpenShift AI Ops Self-Healing Workshop

Welcome to the OpenShift AI Ops Self-Healing Workshop!

OpenShift 4.18+
AI Powered
ML KServe

In this hands-on workshop, you’ll learn how to leverage AI-powered self-healing capabilities in OpenShift. You’ll train machine learning models, interact with your cluster using natural language, and watch automated remediation in action.

What You’ll Learn

Module Description

Module 0

Introduction & Architecture - Understand the hybrid deterministic-AI approach and platform components

Module 1

ML Model Training with Tekton - Train anomaly detection and predictive analytics models using automated pipelines

Module 2

Deploy MCP Server & Configure Lightspeed - Set up the MCP Server and connect OpenShift Lightspeed to your platform (Required)

Module 3

End-to-End Self-Healing - Chat with your cluster using Lightspeed, deploy apps, break things, and watch AI fix them

Module 4

Extra Credit - Advanced ML - LSTM networks, ensemble methods, and deploying your own custom models

Module 5

Notebook Catalog & Use Cases - Comprehensive guide to 33+ notebooks covering all platform capabilities

What’s Already Deployed

Your workshop environment comes pre-configured with:

Component Description

✅ OpenShift Lightspeed

AI assistant integrated into the OpenShift console

✅ MCP Server

Go service connecting Lightspeed to cluster tools

✅ Coordination Engine

Go service orchestrating remediation workflows

✅ KServe ML Models

Anomaly detection + capacity forecasting models

✅ Jupyter Workbench

Notebook environment for ML development

✅ ArgoCD

GitOps deployment via Validated Patterns

Prerequisites

Skills Required

  • Basic OpenShift/Kubernetes knowledge (pods, deployments, services)

  • Familiarity with the OpenShift web console

  • Basic understanding of machine learning concepts (helpful but not required)

Access Information

Your instructor will provide the following credentials:

Verify Your Access

  1. Open the OpenShift Console: https://console-openshift-console.apps.{guid}.example.com

  2. Log in with provided credentials

  3. Click the Lightspeed icon (sparkle icon) in the top-right corner

  4. If the chat interface opens, you’re ready to go!

Workshop Duration

Module Duration Notes

Module 0

15 min

Introduction & overview

Module 1

30 min

ML training pipelines

Module 2

20 min

MCP Server & Lightspeed setup (Required)

Module 3

45 min

Interactive self-healing demo

Module 4

45+ min

Extra credit (Jupyter notebooks)

Module 5

30 min

Notebook catalog reference

Total: ~2.5 hours core (add 75+ min for extra credit and catalog)

Getting Help

If you encounter issues during the workshop:

  1. Check the Troubleshooting sections in each module

  2. Ask your instructor

  3. Review the Troubleshooting Guide

Let’s Get Started!

When you’re ready, proceed to Module 0: Introduction & Architecture to learn how the platform works before diving into hands-on exercises.

Architecture Note: The MCP Server and Coordination Engine are Go services for production performance. The Jupyter notebooks are Python for ML/data science. You don’t need to write Go code to use the platform!