EX380 — the OpenShift exam where you start owning the platform

EX380 earns Red Hat Certified Advanced System Administrator in OpenShift, on OpenShift Container Platform 4.18.

If EX280 is about working competently inside a cluster somebody else runs, EX380 is about being the person who runs it.

Seven objectives, and where ownership starts

• Configure and manage OpenShift authentication and identities
• Back up and restore applications with OpenShift API for Data Protection (OADP)
• Manage workloads with cluster partitioning
• Manage workloads with pod scheduling
• Implement OpenShift GitOps
• Manage cluster monitoring and metrics
• Provision and inspect cluster logging

The one that sinks people

OADP. Backup and restore is the objective candidates most often treat as a reading exercise, and most often fail on.

The concepts are easy. The operations are fiddly: a backup storage location that has to authenticate to object storage, a volume snapshot location, backup schedules, restore ordering, and the interaction between all of those and whatever CSI driver your storage happens to use. Every one of those is a place where a plausible-looking configuration produces a backup that silently contains nothing.

The only preparation that works is the obvious one, and people skip it anyway: take a backup of a real application with real persistent data, delete the namespace, and restore it. If you haven’t watched a restore come back wrong at least once, you don’t know this objective.

That loop needs a cluster you can delete namespaces in, which rules out anything shared. The OpenShift III track on CertLabs, our practice platform, gives you a sandboxed one that resets between exercises.

Cluster partitioning and scheduling

These two sit together in practice. Partitioning is about carving a cluster so workloads land where they should, with infra nodes separated from application nodes and workloads pinned to hardware that suits them. Scheduling is the mechanism: taints and tolerations, node selectors, affinity and anti-affinity, pod topology spread constraints.

The exam-relevant skill is expressing an intent — “these pods must never share a node”, “this workload only runs on nodes with local SSD” — as the right primitive on the first attempt. Know which problems affinity solves and which ones taints solve. They aren’t interchangeable, and reaching for the wrong one produces a configuration that looks correct and schedules wrong.

GitOps

Implementing OpenShift GitOps means Argo CD: applications, projects, sync policies, and the reconciliation model. The mental shift the exam rewards is understanding that you’re no longer applying manifests. You’re describing a desired state and letting a controller converge on it. Candidates who keep thinking imperatively lose time fighting sync waves and pruning behaviour.

Monitoring, metrics and logging

Cluster monitoring on OpenShift is a configured Prometheus stack rather than a blank slate. Know how to enable monitoring for user-defined projects, where the configuration lives, and how to inspect what the stack is scraping. Logging is a separate operator with its own lifecycle.

The framing that helps: the exam doesn’t care whether you can install Prometheus. It cares whether you can operate the one OpenShift ships.

Prerequisites — read this part carefully

Red Hat lists the prerequisite as a current Red Hat Certified Specialist in OpenShift Administration certification or equivalent experience, plus knowledge matching DO180 and DO280 for version 4.18 or later. Earlier 4.x releases are explicitly called out as insufficient.

Take that last clause seriously. OpenShift moves. If your EX280 was several minor versions ago, some of what you know is now wrong in detail. Not wildly wrong, but enough to cost you tasks.

Recommended training is DO280 and DO380 (OpenShift Administration III: Scaling Kubernetes Deployments in the Enterprise).

The renewal argument

Passing EX380 renews your EX280. If you were going to pay to renew anyway, the marginal cost of sitting the higher exam instead is small, and you end up with two live credentials rather than one.

The only case against is if you won’t touch this material at all. But if that’s true, the EX280 renewal probably wasn’t worth buying either.

Who it is for

Red Hat names platform engineers, system administrators, cloud administrators and infrastructure roles on one side; enterprise architects, SREs, DevOps engineers and application roles on the other.

That breadth is a hint about the exam’s character. It sits at the seam between who runs the platform and who ships on it, and the objectives reflect both.