SUSE Edge 3.6: Hands-on Workshop

The Edge Rodeo

Rancher Prime • Elemental • Edge Image Builder • Fleet • Hauler

The scenario¶

Vertex Trust Bank operates ATMs, branch teller terminals, and regional processing hubs across 47 branches. Every edge node must boot from the same OS image, onboard automatically with no hands-on config at the remote end, and be manageable from a single control plane. This workshop deploys the Vertex Trust Bank management stack and walks you through building and booting four edge nodes across two different provisioning paths.

Lab topology¶

KVM host (bare metal)
│
├── rancher   192.168.122.9    Rancher Prime 2.14.1 + Elemental Operator 1.9.0 + Fleet
├── eib       192.168.122.20   EIB 1.3.3.1 + Hauler 1.2.2 + Gitea 1.22
│
├── edge1     192.168.122.31   vTPM 2.0  ──► Elemental onboarding  ──► K3s cluster
├── edge2     192.168.122.32   vTPM 2.0  ──► Elemental onboarding  ──► standalone
├── edge3     192.168.122.33   UEFI      ──► EIB standalone        ──► RKE2 cluster
└── edge4     192.168.122.34   UEFI      ──► EIB standalone        ──► K3s cluster

Two provisioning paths run in parallel throughout this lab:

Elemental path
edge1, edge2
Phone-home onboarding via TPM. The node boots, the Elemental agent contacts the registration endpoint, and the management cluster takes control. Cluster provisioning happens from Rancher.

EIB standalone path
edge3, edge4
The full Kubernetes stack is baked into the disk image. Boot it, you have a running cluster. No phone-home. No registration step.

Exercises¶

Exercise 01

Tour the environment

⏱ 15 min

Check the management cluster, the EIB image factory, and the Hauler artifact store before touching anything.

Exercise 02

Configure Elemental and the network plan

⏱ 30 min

Create the MachineRegistration, get the registration URL, and define a static IP for each node via NMState configs.

Exercise 03

Build four EIB images

⏱ 50 min

One image per node, each with its own static IP baked in. Two Elemental ISOs, one RKE2 RAW, one K3s RAW.

Exercise 04

Seed and boot the four nodes

⏱ 25 min

ISO workflow for Elemental nodes, thin-clone RAW workflow for standalone cluster nodes.

Exercise 05

Provision a K3s cluster

⏱ 20 min

Label a registered Elemental node and watch Rancher provision a full K3s cluster on it remotely.

Exercise 06

Deploy workloads via Fleet

⏱ 20 min

Label clusters for Fleet, import standalone nodes into Rancher, and watch vertex-bank-app deploy across the fleet.

Offline-first design¶

rodeo deploy needs internet access once to pull everything (Rancher, K3s, Elemental, SL Micro images, Helm charts). After that, the lab runs completely offline.

The EIB VM hosts two services that together make the lab self-contained:

Hauler: OCI registry at :5000 (EIB container, elemental-register, vertex-bank-app) and file server at :8080 (SL Micro 6.2 ISO and RAW). EIB pulls container images to embed and the SL Micro base OS from here.
Gitea: local Git server at :3000 with two repos: gitea/vertex-bank-app (Fleet GitRepo source, no GitHub access needed) and gitea/eib-config (EIB image definitions, NMState network configs, and combustion scripts that students clone in Exercise 2).

Edge nodes boot with registries.yaml baked in by EIB, routing all container pulls through Hauler. Fleet syncs from local Gitea. EIB itself pulls its definitions and scripts from Gitea and its binary content from Hauler. The management cluster runs entirely offline after deploy. See the Disconnected environment reference for the full architecture.

Resources¶

📋 Quick reference 🔌 Disconnected environment 🎯 Bonus exercises 📦 Bonus lab: Hauler 📄 Full lab guide (print) 🔧 Instructor: host setup ✅ Instructor: pre-lab checklist ⚙️ rodeo-cli

Component versions¶

Component	Version
Rancher Prime	2.14.1
K3s (management cluster)	v1.35.5+k3s1
cert-manager	v1.20.1
Elemental Operator	1.9.0
Edge Image Builder	1.3.3.1
Hauler	1.2.2
Gitea	1.22
SUSE Linux Micro	6.2