Advanced Troubleshooting for CloudLinux and cPanel

CloudLinux adds per-account limits (LVE) and a caged file system (CageFS) to a cPanel server, which is why most shared hosts use it. When something breaks, it is usually the kernel, the licence, the package repositories or CageFS. This guide gives a server administrator the checks and fixes for each, in the order that finds the cause fastest.

1. Get a baseline first

Before you change anything, record what the server is running. Most "mystery" problems turn out to be a kernel that didn't boot, or a release that is older than you thought.

cat /etc/os-release          # CloudLinux release, e.g. 8.10 or 9.x
uname -r                     # the kernel that is running now
grubby --default-kernel      # the kernel that will boot next time
cldetect --detect-cp         # which control panel CloudLinux detects
uptime                       # when the server last rebooted

If uname -r and grubby --default-kernel disagree, the server was updated but not rebooted, or it fell back to an older kernel on the last boot. Section 2 covers both.

CentOS 7 and CentOS 8 are end of life. A new cPanel server should run a release that cPanel currently supports, such as AlmaLinux, Rocky Linux, CloudLinux or Ubuntu LTS; the list changes between cPanel versions, so check cPanel's system requirements. See why AlmaLinux and Rocky Linux replaced CentOS.

2. Kernel and LVE checks

LVE is the kernel feature that enforces CPU, memory, process and I/O limits for each account. If it isn't loaded, limits silently stop working.

So on CloudLinux 9, a kernel name without "lve" is normal. Check that LVE is actually working instead:

lsmod | grep -i lve          # the LVE module is loaded
lvectl list | head           # accounts and their limits
lveps                        # current usage per account
systemctl status lve lvestats

lvectl and lveps are command-line tools, not services, so there is no systemctl status lvectl to check.

Boot the right kernel

rpm -qa 'kernel*' | sort                   # installed kernels
grubby --info=ALL | grep -E '^(index|kernel)'
grubby --set-default /boot/vmlinuz-<version>
reboot

After the reboot, run uname -r again. Keep at least one older kernel installed as a fallback, and don't remove the running one.

Kernel updates without a reboot

If KernelCare is installed, it patches the running kernel live. kcarectl --info shows whether it is active, and kcarectl --uname shows the effective patched version. A full kernel update with dnf update kernel* still needs a reboot to take effect.

3. Licence, registration and repositories

CloudLinux packages come from the CloudLinux Network, and the server must hold a valid licence to reach it. Errors such as "This system is not registered", repository 403s or missing packages usually trace back here, not to a broken .repo file.

Don't paste a repository file from a forum over the one CloudLinux installed. If the release package is damaged, reinstall it from its own repository (for example, dnf reinstall cloudlinux-release) and let it write the correct configuration.

4. CageFS problems

CageFS gives every user a private view of the file system. When a site suddenly can't find a binary, a PHP extension or a newly installed tool, the cage is often out of date.

cagefsctl --display-user-mode      # is CageFS on for all users, or opt-in?
cagefsctl --list-disabled          # users outside the cage
cagefsctl --force-update           # rebuild the cage after installing software
cagefsctl --remount-all            # remount every user's cage

To test whether a problem is caused by the cage, disable it for one user with cagefsctl --disable username, retry, and enable it again with cagefsctl --enable username. Leave it disabled only as long as the test takes; outside the cage, that user can see far more of the server.

5. CloudLinux Manager errors

On a cPanel server, CloudLinux Manager is the WHM plugin (package lvemanager) where you set LVE limits, the PHP and Node.js selectors, and CageFS options. If it shows a blank page or a Python traceback such as ModuleNotFoundError:

• Don't run pip install to add the missing module. The tools use CloudLinux's own Python packages, and pip changes files that dnf manages, which breaks the next update.
• Do reinstall the packages that provide it: dnf reinstall lvemanager lve-utils, then reload WHM.
• If cldiag is available on your release, cldiag --all runs CloudLinux's own consistency checks and names the failing component.
• If the error survives a reinstall, send the full traceback to CloudLinux support with your licence details.

6. cPanel's side of the problem

Some faults that look like CloudLinux are really cPanel packages or configuration.

/usr/local/cpanel/scripts/check_cpanel_pkgs --fix   # repair cPanel-managed RPMs
/usr/local/cpanel/scripts/upcp                      # run a cPanel update
/usr/local/cpanel/scripts/restartsrv_httpd          # restart Apache the cPanel way
tail -100 /usr/local/cpanel/logs/error_log          # cPanel's own errors

For PHP builds and Apache modules, use EasyApache 4 rather than installing packages by hand; see EasyApache 4 with CloudLinux. If you need to hold a package at one version, use a version lock as described in managing dnf version locks on CloudLinux.

7. Where the evidence is

Change one thing at a time, and write down what you changed. Take a snapshot or backup before kernel and release upgrades.

8. Where Domain India fits

If you'd rather not manage CloudLinux yourself, our shared cPanel hosting runs on CloudLinux with CageFS, and we look after the server, security updates and weekly backups. Prices on the card are live and exclude 18% GST.

If you want root access, a Domain India VPS is self-managed. cPanel is not offered on our VPS plans; you can install cPanel and CloudLinux yourself with licences you buy elsewhere, as described in installing cPanel on a VPS, or pick one of the panels offered at checkout.

Ready to decide? Compare cPanel shared hosting with a self-managed VPS, or open a ticket if a limit or error on your shared account needs a look.