Red Hat Exam Day Protocol
The loop to run on every task in a performance-based Red Hat exam - the persistence checklist, time management, the reboot ritual, and the nine mistakes that turn a working configuration into a zero.
Applies to any performance-based Red Hat exam. The loop below is identical for
each; only the allowed documentation differs — the core RHEL exam gives you man
and /usr/share/doc, the Ansible one adds ansible-doc.
Part of RHCSA to RHCE - A 60-Day Lab Curriculum.
The loop
Read requirement
↓
Identify hosts ← which group? one host or all?
↓
Implement
↓
Validate
↓
REBOOT
↓
Validate again
That last step is not optional. Red Hat exams are performance based and graded on the state of the machine after a reboot. A configuration that works now but disappears on reboot is not a partial answer — it is a zero.
The persistence checklist
Before you consider any task finished, ask which of these applies:
| Did you… | Persistence requirement |
|---|---|
| Start a service | systemctl enable as well |
| Mount a filesystem | Entry in /etc/fstab, by UUID |
| Set a SELinux boolean | setsebool -P |
| Set a SELinux context | semanage fcontext + restorecon, not chcon |
| Open a firewall port | --permanent and --reload |
| Configure an IP | An nmcli profile with autoconnect, not ip addr add |
| Set a hostname | hostnamectl, not hostname |
| Add swap | An fstab entry |
| Change a kernel parameter | /etc/sysctl.d/, not just sysctl -w |
| Create a timer or cron job | Enabled, and Persistent=true for timers |
| Run a rootless container | loginctl enable-linger, or it never starts at boot |
findmnt --verifybefore every reboot. A malformed/etc/fstabdrops the machine into emergency mode and can burn twenty minutes of exam time.findmnt --verifyandmount -acost three seconds.
Allowed documentation
Learn to move fast in these. They are all you get.
# Ansible exam only
ansible-doc -l | grep -i <thing> # find the module
ansible-doc <module> # full options
ansible-doc -s <module> # paste-ready skeleton ← use this most
ansible-galaxy collection list # what is installed
# Any exam
man <command>
man -k <keyword> # apropos, when you forget the name
<command> --help
ls /usr/share/doc/ # config examples live here
/usr/share/doc is badly underused. Example configurations for httpd, chrony,
autofs and others are sitting right there, ready to copy.
Time management
First 10 min Read every task. Number them by difficulty.
Do NOT start on task 1 by default.
Then Easy, high-confidence tasks first. Bank the marks.
Middle Hard tasks, with a hard time cap on each.
Last 30 min REBOOT. Then verify every task from the top.
- Never let one task eat more than about 15% of your time. Flag it, move on, come back.
- Tasks are independent. A task you cannot do does not block the rest.
- If a task depends on an earlier one you failed, do it anyway — partial state may still score.
The reboot ritual
Reboot at least twice: once in the middle, once near the end. A mid-exam reboot surfaces persistence failures while you still have time to fix them. Finding them in the last five minutes is finding them too late.
findmnt --verify # fstab sane?
mount -a # no errors?
systemctl --failed # nothing failed?
reboot
# after it returns:
systemctl --failed
df -h; swapon --show; getenforce
firewall-cmd --list-all
# then re-verify every task
The nine mistakes that turn work into a zero
- Service started but not enabled.
setseboolwithout-P.chconinstead ofsemanage fcontext+restorecon— survives a reboot, dies on a relabel.- A firewall rule left runtime-only, with no
--permanent. - fstab using
/dev/sdb1instead of a UUID, after device names shift. setenforce 0to "make it work" — SELinux must be enforcing.- A playbook that reports success but
skipped=everything. - A plaintext password where a hash was required.
- Reading the requirement as you assume it, not as written. Read it twice.
Number 6 deserves emphasis: disabling SELinux to unblock yourself converts a partial score into a guaranteed zero across multiple tasks, because the graders check enforcement separately from the thing you were trying to fix.
If you get stuck
- Re-read the requirement. Half of all stuck moments are misread requirements.
ansible-doc -sorman -k. The answer is on the machine.- Check the logs:
journalctl -xeu <unit>,ausearch -m AVC -ts recent. - Still stuck after your time cap? Flag it, move on, come back.
Never sit frozen, and never disable a security control to unblock yourself.
The night before
- No new topics. Cramming something unfamiliar displaces something you know.
- Confirm the exam time, ID requirements and the check-in process.
- Verify your machine and environment requirements if it is a remote exam.
- Sleep properly. Tired beats under-prepared in the wrong direction.
Practising under these conditions
None of the above works as advice; it only works as habit. The way to build the habit is to sit full timed mocks on a machine you have deliberately broken beforehand:
- Two RHCSA Mock Exam Papers — two 2.5-hour papers, plus a hostile variant
- Ansible Scenario Labs and the EX294 Mock — the four-hour Ansible paper
- Verifying Your Own Labs Offline — a grader that scores only what survived a reboot
- Ten Advanced RHEL Break-Fix Drills — for the troubleshooting half