Operating Applications on OpenShift 5: Platform Engineering, AI & GitOps

Work through the responsibilities of a platform engineer: establish an application baseline, understand GitOps, apply tenant and resource controls, investigate an incident, and deliver a verified fix.

Module 1 is Mission National Parks: a 25–30-minute console-first workshop (28-minute guided path). Students deploy the frontend, Python API and working MongoDB 8, encounter an incorrect MongoDB Service target port. Lightspeed 2.0 helps diagnose the connection failure; students repair one Service field, load data and verify the map.

The workshop is designed for up to 30 participants on a shared OpenShift cluster. Each participant uses an assigned identity and application namespace.

Your lab credentials
Item Value

Username

user-guid

Password

password

Workshop project

user-guid-demo

OpenShift console

Open the console

Try the OCP Console tab too, it may suit you better.

Your login username is user-guid. For Module 1, select project user-guid-demo. Before running any terminal commands, log in to oc as user-guid using the OpenShift console’s Copy login command. The default Showroom terminal identity may not have permission to patch your Deployment.

Module 1 uses the existing user-guid-demo project. Students create their database Secret and deploy three components without changing the shared Argo CD and Gitea services. Some screenshots may need updating for the MongoDB Service-port exercise.

This workshop targets OpenShift 5.0.0 and OpenShift Lightspeed 2.0. The actual cluster release and installed Lightspeed version come from external provisioning, not this Showroom attribute. Instructors must verify both on the live cluster. OpenShift 5.0 documents Kubernetes 1.36 and general availability of the MCP server for Red Hat OpenShift; optional toolsets and agentic integrations have separate requirements and support statuses.

What You Will Do

  • Select your existing user-guid-demo project, then explore the updated OpenShift 5 Project Felt console. Do not create a namespace.

  • Deploy parksmap from a container image using the native console’s Quick create flow.

  • Import the Python nationalparks API from Git and observe its build in Topology.

  • Create a MongoDB Secret with your own passwords in your assigned project.

  • Import a working MongoDB 8 Deployment, bootstrap ConfigMap and internal Service through the console YAML editor.

  • Connect the backend to your Secret and observe the deliberately incorrect MongoDB Service target port.

  • Use Lightspeed to investigate the MongoDB connection failure, repair its Service port and load park data.

  • Complete a workload-readiness card and continue into GitOps onboarding.

This module deliberately distinguishes new OpenShift 5 appearance/MCP capabilities from long-standing Deployments, Routes, Builds, YAML import and GitOps.

Workshop Structure

Module Duration Activity and outcome Content status

1 — Mission National Parks

25–30 min (rehearsal target)

Native console deployment: frontend, Git backend, student-created Secret, working MongoDB 8, broken MongoDB Service target port and Lightspeed-guided repair.

Console deployment draft; test images/builds and recapture screenshots

2 — GitOps: National Parks Full Stack

45 min (rehearsal target)

Use the GA GitOps 1.22 console plug-in and Gitea to deploy the complete ParksMap, Python and MongoDB stack with Auto-Sync and Self-Heal; observe Builds, Routes, dataset loading, a Git commit and tenant isolation.

Main full-stack Auto-Sync lab; requires preflight of student Gitea copies, image builds, PVCs and GitOps permissions

3 — Agentic Application Operations

40 min

Introduce a controlled ParksMap readiness incident through student Gitea; use Lightspeed and (only where secure per-student sandbox isolation is enabled) a student-owned analysis-only AgenticRun; otherwise instructor demo. Repair through GitOps.

Five-mission hands-on lab; AgenticRun demonstration requires separate Agentic Operator and RBAC preflight; read-only evidence path available when not installed

4 — WeatherNow Agentic Troubleshooting

40 min

Deploy lightweight WeatherNow as a separate GitOps Application; introduce a bad readiness probe in student Gitea, investigate with Lightspeed or an authorized student-owned AgenticRun, then restore the source.

WeatherNow uses only 200m CPU and 128Mi memory limits; no database or PVC. AgenticRun student creation defaults disabled until operator sandbox isolation is proven.

5 — WeatherNow GitOps Release Investigation

30 min

Introduce an invalid WeatherNow image tag in student Gitea, inspect ImagePullBackOff through Lightspeed or an approved analysis-only AgenticRun, and repair GitOps source.

Fully separate incident; source-controlled recovery without editing National Parks or cluster-wide settings

Planned full workshop: approximately 3 hours 5 minutes (pending live rehearsal of AgenticRuns and WeatherNow Modules 4–5). Module 1 targets 28 guided minutes, allowing up to 30 minutes in a preflighted student environment. Registry pulls, Git builds, student Secret creation, discovery RBAC, the deliberate MongoDB Service port fault, Lightspeed 2.0 reads and the console repair must be rehearsed to meet this timing. The former scaling and ingress-controller inspection exercises are outside this timebox. Module 2 now deploys an independent, full National Parks stack to each student’s -gitops namespace with automatic reconciliation; Module 1 remains unchanged in -demo. Module 3 uses a deliberately incorrect GitOps-managed ParksMap readiness port to demonstrate why Synced does not imply Healthy, followed by evidence-backed Lightspeed and optional analysis-only AgenticRun investigation; it ends with both stacks healthy again.

Release Topics Connected to the Application

Topic Workshop connection

Project Felt console design

Experience the updated OpenShift 5 visual style while using familiar navigation. In Module 2, the GitOps console plug-in is required; console appearance themes may be explored when available.

Kubernetes and runtime foundation

Compare the actual platform release with the documented Kubernetes 1.36 / CRI-O foundation for OpenShift 5.0.

MCP server — GA in 5.0

Use Lightspeed 2.0 Troubleshooting to inspect MongoDB Services, Pods, EndpointSlices and backend logs. Students apply a one-field Service port fix.

HAProxy version selection

Check normal ingress behaviour and identify the serving router’s status.effectiveHAProxyVersion: 3.2 by default, or the supported 2.8 migration option.

Mirrored image credentials — GA

Inspect application image references and explain the administrator-configured project-scoped credential path for mirrors.

Agentic update assessment — Technology Preview

An optional instructor demonstration uses the Lightspeed-driven Cluster Update page. The Agentic Operator is a separate prerequisite.

Agentic application troubleshooting

Module 3 investigates National Parks; Modules 4–5 use lightweight WeatherNow. Student-created analysis-only AgenticRuns remain disabled unless tenant sandbox isolation and admission have passed admin preflight; otherwise Lightspeed and instructor demonstration remain available.

Extension lifecycle

Identify the components providing workshop capabilities and check their individual versions and support status.

Unified console navigation predates OpenShift 5. The core Module 1 exercise uses the already configured Lightspeed MCP connection. The gateway/Lifecycle Operator deployment path and the agentic update-assessment integration are separate setups and are not core-lab prerequisites.

Your Environment

Component Preparation required

Console and terminal

Log in with user-guid / password using the console link above. Module 1 uses the native console; the instructor and later modules also need oc, curl, jq, git, and python3.

Assigned workshop project

user-guid-demo — an existing project which may already host Argo CD, Gitea and Showroom. No National Parks workloads should exist yet. The instructor must verify Secret-creation and discovery permissions in this exact namespace.

National Parks

Not pre-deployed. The instructor prepares namespace-scoped discovery RBAC and validates the frontend image, Python source builder, MongoDB image and students' permission to create Secrets. Students create their own database Secret, deploy the three components and initialize disposable data.

Application grouping

Workload and pod labels identify the three National Parks components together.

AI integration

Lightspeed Operator 2.0 with live MCP read tools. Students manually repair the MongoDB Service target port; write tools are not required.

Observability

Permission to inspect application logs and workload status; additional metrics are used in later modules.

GitOps

Tenant bootstrap imports an editable openshift-5-container copy into each student’s Gitea. Module 2 creates an auto-syncing Argo CD Application from the root Kustomization, deploying MongoDB, Python API, ParksMap and their support resources into a separate -gitops namespace.

OpenShift console: Open the console

Try the OCP Console tab too, it may suit you better.

If this link still contains %GUID% or example.com, use the URL supplied with your workshop credentials.

Before You Start

This workshop is intended for intermediate platform engineers and site reliability engineers. You should be comfortable with pods, Deployments, namespaces, basic RBAC, and running oc commands. Later exercises also require basic Git operations.

Sign in as user-guid in both the console and terminal. Never use the default Showroom ServiceAccount for workshop commands. In the Showroom Terminal tab, paste this block and confirm the second command prints user-guid before patching anything in user-guid-demo:

oc login --server=https://api.cluster-%GUID%.example.com:6443 --username=user-guid --password='password' --insecure-skip-tls-verify
oc whoami

Module 1 starts in user-guid-demo. Students deploy the frontend, Python backend and working MongoDB 8, investigate the preconfigured MongoDB port fault with Lightspeed, repair it in the console, then load park data once.

Reference Material

This revision uses the OpenShift 5.0 documentation source reviewed on 4 November 2026. See the instructor notes for the pinned source revision and the distinction between the core MCP server, optional toolsets, and separate integrations. Recheck the installed build before delivery.