Advanced Cluster Management Workshop

Overview

This workshop provides hands-on exercises for Red Hat Advanced Cluster Management (RHACM) 2.15 on OpenShift 4.20.

You will learn to manage a multicluster fleet, deploy applications with ArgoCD, enforce governance policies, and integrate AI-powered operations — all from a single ACM hub.

Target versions: OpenShift 4.20 / RHACM 2.15

Base Environment

This workshop was built on top of the Advanced Cluster Management for Kubernetes Demo from the Red Hat Demo Platform. Order this catalog item to get a pre-configured hub cluster with ACM already installed, then follow the modules below.

Prerequisites

A running OpenShift 4.20+ cluster with RHACM 2.15 installed (see Base Environment above)
The oc CLI tool installed (the setup script will attempt to install it if missing)
The kubectl CLI tool installed (the setup script will attempt to install it if missing)
The git CLI tool installed
A GitHub account
AWS credentials (for provisioning managed clusters via Hive)

Platform Compatibility

Platform Status

Platform	Status
RHEL 9 (Red Hat Demo Platform bastion)	Fully tested
macOS	Supported, not fully tested. Scripts detect your OS and install tools via Homebrew. Some module exercises use GNU `sed -i` syntax — you may need `brew install gnu-sed` and use `gsed` in place of `sed`.
Windows	Not supported. Use WSL2 or the RHEL bastion.

RHEL 9 (Red Hat Demo Platform bastion)

Fully tested

macOS

Supported, not fully tested. Scripts detect your OS and install tools via Homebrew. Some module exercises use GNU sed -i syntax — you may need brew install gnu-sed and use gsed in place of sed.

Windows

Not supported. Use WSL2 or the RHEL bastion.

Quick Start

Clone the repository and run the setup validation script:

git clone https://github.com/tosin2013/rhacm-workshop.git
cd rhacm-workshop
./setup.sh

The setup script will:

Detect your platform (Linux or macOS) and check for required CLI tools (oc, kubectl, git).
Attempt to auto-install any missing tools via dnf (RHEL/Fedora) or brew (macOS).
Validate your OpenShift cluster, ACM installation, managed clusters, storage, and APIs.
Check AWS credentials if provisioning managed clusters (skip with --skip-aws).

Workshop Architecture

Hub Cluster (OCP 4.20, ACM 2.15, SNO)
  ├── local-cluster   (self-managed, labels: environment=hub)
  ├── standard-cluster (Hive AWS SNO, labels: environment=dev)
  └── gpu-cluster      (Hive AWS SNO g6.4xlarge, labels: gpu=true, accelerator=nvidia-l4)

Troubleshooting

"Too many pods" / FailedScheduling

If you see pods stuck in Pending with an event like 0/1 nodes are available: 1 Too many pods, the node has hit the Kubernetes maxPods limit (default 250 on SNO clusters). This commonly happens during Exercise 3 (Observability) or Exercise 4 (ArgoCD) when many pods are deployed.

Fix: Run the included script to raise the limit to 350. This creates a KubeletConfig and triggers a node reboot (~10-15 minutes):

bash scripts/bump-max-pods.sh 350

Monitor the rollout with oc get mcp master -w. After the node returns, pending pods will schedule automatically.

switch-to-grafana-admin.sh errors (Exercise 3)

This script lives in the multicluster-observability-operator repo, not this workshop repo. You must:

Clone the repo: git clone --depth 1 https://github.com/stolostron/multicluster-observability-operator.git
Log into the Dev Grafana URL in your browser first (the script looks up your user in the Grafana database — if you haven’t logged in, the user doesn’t exist yet)
Run from the tools directory: cd multicluster-observability-operator/tools && bash switch-to-grafana-admin.sh kube:admin