Free EX342 Practice Questions

Question 1

A reporting server with eight logical CPUs slows down during a batch run. The RHEL web console and command-line tools show the following throughout the same five-minute interval:

Load average: 11.8, 9.4, 4.1
vmstat: r = 1-2, b = 9-12, si = 0, so = 0
CPU: 68% idle, 22% I/O wait
Storage device await: 130 ms; its usual value under this workload is 3 ms
Most stalled workers have process state D.

Where should the administrator concentrate the investigation?

Accepted Answer

Blocked I/O: correlate the waiting workers with device latency and kernel storage messages.

Answer

CPU contention: examine worker affinity and CPU scheduling.

Answer

Active swapping: inspect swap traffic and memory pressure on the host.

Answer

Zombie accumulation: investigate unreaped children and the parent process that should collect their exit status.

Question 2

Apache serves static files from /srv/catalog on a RHEL host in SELinux enforcing mode. An administrator previously made the site work with chcon, but access fails again after a filesystem relabel. An AVC records httpd_t being denied file read access to var_t content. UNIX permissions and Apache's access configuration are correct, and no local file-context rule exists for this subtree.

Select the repair that survives future relabeling and grants the web process SELinux read access without granting it write access to the content.

Accepted Answer

semanage fcontext -a -t httpd_sys_content_t '/srv/catalog(/.*)?'; restorecon -Rv /srv/catalog

Answer

chcon -R -t httpd_sys_content_t /srv/catalog; systemctl restart httpd.service

Answer

setsebool -P httpd_can_network_connect on; systemctl restart httpd.service

Answer

semanage fcontext -a -t httpd_sys_rw_content_t '/srv/catalog(/.*)?'; restorecon -Rv /srv/catalog

Question 3

An API fails at boot because its database is not started. The only dependency relationship in api.service is After=report-db.service. The database unit is disabled, is not masked, and has no other activation trigger. It uses Type=notify and sends READY=1 only when it can accept connections.

Starting api.service must also start the database, and a failed database activation must prevent the API from starting. Which additional directive belongs in api.service's [Unit] section?

Accepted Answer

Requires=report-db.service

Answer

Before=report-db.service

Answer

Wants=report-db.service

Answer

PartOf=report-db.service

Question 4

A post-change audit runs rpm -V rsyslog after an approved edit to its forwarding destination:

S.5....T. c /etc/rsyslog.conf

The configuration diff matches the authorized edit, and end-to-end forwarding tests pass. No other verification differences are reported. What disposition fits this finding?

Accepted Answer

Record the expected size, digest, and modification-time differences alongside the approved change.

Answer

Restore packaged permissions and ownership, then repeat verification.

Answer

Rebuild the RPM database so approved changes no longer appear in file verification.

Answer

Quarantine the host until the changed package-owned configuration passes incident-response review.

Question 5

A service's health check succeeds at http://127.0.0.1:8080/health on server 192.0.2.40, but clients cannot connect to 192.0.2.40:8080. The server capture shows each client SYN arriving and a reset leaving in reply. Its only listener on that port is:

LISTEN 0 128 127.0.0.1:8080 0.0.0.0:*

The active firewall zone permits TCP 8080, and routing between the hosts is correct. The service is intended to accept connections on the server's LAN address. What needs to change?

Accepted Answer

Change the application's listening address from 127.0.0.1 to 192.0.2.40, then restart it.

Answer

Reload the active firewall zone after adding TCP 8080 permanently.

Answer

Install an explicit return route for the remote client subnet.

Answer

Increase the application's connection backlog, then restart it to accept additional pending client connections.

Question 6

An XFS archive volume shuts down after metadata I/O errors. The storage controller reports new uncorrectable media errors on the same underlying device, and reads from several affected regions fail. Application writes have been stopped. Healthy storage large enough to hold a recovery image is available.

Before anyone runs xfs_repair on the original volume, what is the priority?

Accepted Answer

Preserve recoverable data on healthy storage and address the failing device before filesystem repair.

Answer

Unmount the volume and repair its metadata on the original device.

Answer

Unmount the volume and use xfs_repair -L to clear the damaged log.

Answer

Reboot into rescue mode and replay the original volume's log before replacing the device.

Question 7

In the tasks section of an Ansible play, a task copies a changed rsyslog forwarding configuration and notifies a handler named Restart rsyslog. The next task successfully validates the configuration with rsyslogd -N1. A following logger task sends a test message, but that message still follows the old configuration; the handler runs only after the remaining tasks finish. The receiver and network path have already been verified.

The test must use the new configuration, while runs with unchanged configuration must not restart rsyslog. How should the play be adjusted?

Accepted Answer

Insert ansible.builtin.meta: flush_handlers after validation and before the logger task.

Answer

Set force_handlers: true for the play and retain the existing task order so notification is honored despite subsequent task failures.

Answer

Insert a systemd_service task with state: restarted before the logger task, with no condition on whether the configuration changed.

Answer

Add enabled: true to the restart handler and leave the logger task before the point at which pending handlers execute.

Question 8

A trace accounts for every module in a login service's PAM authentication stack:

1. A required pam_faillock.so check returns failure.
2. A sufficient pam_unix.so check validates the supplied password.
3. All remaining authentication checks return success.

There are no include, substack, or extended-control jumps affecting this sequence. The application denies the login. Why did successful password verification not produce a successful authentication result?

Accepted Answer

A sufficient success cannot override an earlier required failure; the authentication stack still fails.

Answer

The required failure prevented any subsequent authentication module from executing.

Answer

The sufficient success reset the accumulated result, so rejection occurred elsewhere.

Answer

A successful sufficient module is accepted only when a later required module independently validates the password.

Question 9

A storage change must leave /dev/vgdata/archive at exactly 60 GiB, with its mounted XFS filesystem at /srv/archive expanded to use the LV. The LV and filesystem were both 48 GiB before the change. An earlier step extended the LV, but the filesystem-growth step was skipped. Current LVM output shows:

Extent size: 8 MiB
Current LV extents: 6656
Free VG extents: 8192

The LV is thick and linear; the filesystem is healthy and supports online growth. Which remaining operation meets the change request?

Accepted Answer

Run lvextend -l +1024 /dev/vgdata/archive, then xfs_growfs /srv/archive.

Answer

Add 1536 extents with lvextend -l +1536 /dev/vgdata/archive, then grow XFS with xfs_growfs /srv/archive.

Answer

Add 7680 extents with lvextend -l +7680 /dev/vgdata/archive, then grow XFS with xfs_growfs /srv/archive.

Answer

Run xfs_growfs /srv/archive without extending the LV again, allowing the filesystem to consume the earlier allocation.

Question 10

After a virtual machine moves to VirtIO block storage, boot stops in the initramfs because the root disk is not discovered. Rescue media can read the disk, and its UUID matches the installed boot entry. The installed kernel has virtio_blk available, but lsinitrd shows it missing from that kernel's initramfs; the required supporting drivers are present.

You are inside a fully prepared chroot of the installed system, with its /boot mounted. KVER contains the exact installed kernel version to repair; uname -r still identifies a different rescue kernel. Choose the repair that targets the failing boot image.

Accepted Answer

Run dracut --force --add-drivers virtio_blk /boot/initramfs-${KVER}.img "$KVER" inside the installed-system chroot.

Answer

Regenerate the GRUB menu from the installed configuration, leaving the existing initramfs in place because the root UUID is correct.

Answer

Load virtio_blk into the running rescue kernel with modprobe, confirm the disk appears, and reboot without rebuilding the installed image.

Answer

Run dracut --force /boot/initramfs-$(uname -r).img "$(uname -r)" inside the chroot, then retry the original installed boot entry.

Free EX342 Practice Questions

Domain 1: Understand and employ general methods for troubleshooting

Question 1

Question 2

Domain 2: Diagnose and troubleshoot system startup issues

Question 3

Question 4

Domain 3: Diagnose and troubleshoot file system issues

Question 5

Question 6

Domain 4: Resolve package management issues

Question 7

Domain 5: Troubleshoot and fix network connectivity issues

Question 8

Domain 6: Diagnose application issues

Question 9

Domain 7: Identify and fix authentication issues

Question 10

The rest of the EX342 blueprint

That's 10 of 1,030