Took the Red Hat EX188 (Containers with Podman). Here's what nobody mentions about Red Hat's performance-based format.

Posted by Other-Competition-86@reddit | sysadmin | View on Reddit | 7 comments

For anyone eyeing RHCSA, RHCE, or any of the EX-series — the format is radically different from every cloud cert and most people show up unprepared. Sharing what bit me so you can avoid it.

The format is "do the thing", not "pick the right answer". You get a live RHEL system over a remote KVM session and a list of tasks. Build this image. Run this rootless container with these mounts. Set up systemd units to auto-restart it. The grader runs scripts against the box at the end. There is no answer key — your filesystem state IS the answer.

Things the prep books undersell:

1.⁠ ⁠You will lose 10+ minutes to muscle memory. Type ⁠ docker ⁠ instead of ⁠ podman ⁠, hit enter, get "command not found", repeat. Train it out before exam day. Alias it during practice if you have to.

2.⁠ ⁠Rootless containers are not a footnote — they're 30% of the exam. UID mapping, ⁠ ~/.config/containers/ ⁠, ⁠ --userns=keep-id ⁠, and the joys of ⁠ subuid ⁠/⁠ subgid ⁠. If your prep just shows you ⁠ podman run --rm -it alpine sh ⁠, you're studying for half the exam.

3.⁠ ⁠Containerfile, not Dockerfile. Both work. They expect you to use the Red Hat name. Same with ⁠ registry.access.redhat.com/ubi9/* ⁠ over docker.io images — the box has no internet.

4.⁠ ⁠Systemd integration is the part everyone forgets. ⁠ podman generate systemd --new --name foo ⁠ and Quadlets (⁠ .container ⁠ files in ⁠ /etc/containers/systemd/ ⁠). Quadlets are the new way and they show up. Don't assume "I'll figure it out on the day".

5.⁠ ⁠Time pressure is real. ~2 hours, 8-12 tasks. If you stare at one task for >15 min, move on and come back. Partial credit exists. Zero on a task you didn't touch is the same as a wrong answer, but a half-finished container at least shows attempt.

6.⁠ ⁠There's no ⁠ man ⁠ page for the gotchas. SELinux contexts on bind mounts (⁠ :Z ⁠ vs ⁠ :z ⁠), volumes vs binds, why your ⁠ --restart=always ⁠ flag is silently ignored under a rootless service — the kind of stuff that takes one production incident to internalize. Practice in a VM until your fingers know it.

What I wish I'd done:

•⁠ ⁠Spent more time on ⁠ podman pod ⁠ (yes, it does pods, no it isn't k8s) — small but reliable section •⁠ ⁠Practiced building images that fail and reading ⁠ podman build --log-level=debug ⁠ •⁠ ⁠Drilled ⁠ podman play kube ⁠ because they like to mix in a small kube-yaml-to-podman task

TL;DR for sysadmins thinking about it: if you've been running Docker for years and Red Hat keeps poking you about Podman, this is a useful forcing function. The exam rewards muscle memory and operational habits, not memorization. Your day job prepares you for this better than any course does.