Subject: System-wide corruption and "0 bytes" binaries after disk filled to 100%
Hello,
I am facing a critical issue on my server (AlmaLinux 8 / cPanel v114). The disk reached 100% usage, which resulted in widespread file corruption within the cPanel directory structure.
Current Situation:
-
SSH Access: Intermittent/Lost after reboot.
-
WHM Access:
ERR_CONNECTION_REFUSED(Port 2087). -
Mail Services: Completely down. Users receive "Socket errors".
-
Binaries: Several core binaries, including
/usr/local/cpanel/cpsrvd, were found to be 0 bytes in size.
Steps Taken So Far:
-
Disk Space: Freed up space; current usage is well below limits.
-
Service Restarts: Attempted
/scripts/restartsrv_cpsrvdandrestartsrv_cpdoveauthd, but receivedNo such file or directory. It appears the/scriptssymlink or directory is missing/corrupted. -
Update Attempts:
-
Ran
/usr/local/cpanel/scripts/upcp --force. It reported "Maintenance complete", but many files are still missing. -
Attempted
sh /root/cpupdate.sh --force. It failed with the error: "You must install cPanel & WHM on a clean server".
-
-
Service Status:
systemctl restart cpdoveauthdreturnsUnit not found. Manual execution of/usr/local/cpanel/cpsrvd --debugyields no output (due to the 0-byte file corruption). -
Firewall: Disabled
csf(csf -x) to rule out network blocking, but connection is still refused.
Specific Errors:
-
-bash: /scripts/restartsrv_cpdoveauthd: No such file or directory. -
(XID cknxf6) The “cpdoveauthd” service is not a known service. -
ls -lh /usr/local/cpanel/cpsrvdshows-rwx------ 1 root root 0 May 4 04:41 /usr/local/cpanel/cpsrvd.
Request: How can I force a complete re-installation of all cPanel binaries and scripts without performing a clean OS install, as I need to preserve client data? Is there a way to override the "clean server" check in the installer to repair the current corrupted environment?
-
Hey there! I'm not seeing a "
/scripts/restartsrv_cpdoveauthd" on a test system so it seems that missing is normal behavior.However, not having the cpsrvd script isn't normal.
I would try running /scripts/check_cpanel_pkgs to see if that helps get missing tools installed, and then do an additional upcp after that. Can you try that and let me know the results?
0 -
Dear cPRex and cPanel Support Team,
I am providing an update and requesting a definitive migration strategy for my agency. I need to move from a "repair" attempt to a "migration" workflow immediately.
- Current Status of the Corrupted Server
The original server (AlmaLinux 8 / cPanel v114) experienced 100% disk usage, causing severe binary corruption:
- The Problem: Core cPanel binaries (like /usr/local/cpanel/cpsrvd) are 0 bytes.
- Access Issues: SSH is broken/denied, and WHM is unreachable. chroot via Rescue System fails with a Segmentation fault.
- Active Services: Despite the cPanel corruption, the websites are currently online, and the MySQL service is still running and serving data.
- New Environment Ready
- Drive A (Corrupted/Active): Original NVMe with all client data, running LiteSpeed and MySQL.
- Drive B (Fresh): A second NVMe with a clean installation of AlmaLinux 9. cPanel is not yet installed.
- Critical Questions for Migration
Given that the corrupted OS cannot be managed via WHM or SSH, but the sites and MySQL are still active:
- Transfer Tool via Local Mount: Can I install cPanel on the fresh AlmaLinux 9 drive and use the Transfer Tool to migrate accounts directly from the corrupted drive if I mount it locally?
- Live MySQL Migration: What is the best practice for migrating MySQL databases from a drive where the OS is corrupted but the MySQL service is currently active? Should I attempt a mysqldump from the Rescue System or copy the data directory?
- LiteSpeed Preservation: Since the old server uses LiteSpeed, what is the correct sequence to ensure these settings migrate correctly to the new AlmaLinux 9 environment?
Please provide a definitive recommendation. I cannot risk further downtime by attempting to fix a "clinically dead" OS. I need to migrate to the new drive now.
0 -
I'm sorry to hear about the situation. I'll make sure to answer these questions in order to ensure we don't miss anything.
1 - No, there isn't a way to do this with any tools provided by cPanel. The fastest way to get things working would be to restore from backups. While you could perform a manual restoration from the corrupted drive it would be a process of manually creating each cPanel account and manually moving the data to the correct areas - a very tedious job that is both error prone and slow.
2 - Since this isn't work that cPanel handles I don't have a recommendation on this one. We just don't have tools for this, so I can't tell you what the best option is. Ultimately, restoring from a backup would ensure the data isn't corrupted.
3 - It would depend which settings you're talking about here. As LiteSpeed is a drop-in replacement on cPanel systems, many of the settings would be standard Apache values, which you may be able to find in /etc/apache2/conf/httpd.conf. LiteSpeed-specific values that are *not* handled by Apache can be found in /usr/local/lsws/conf/httpd_config.xml
I'm sorry I don't have better options for you, although I still think it would be worth a ticket to have an admin check the cPanel installation to see if it can be resolved. Running a normal "yum update" to confirm the operating system itself is working and then following up with "/scripts/check_cpanel_pkgs" would tell you a lot about the state of the machine.
0 -
I am experiencing a fatal error while installing cPanel on a fresh AlmaLinux 9 server (Hetzner).
The main issues are:
-
Installation Error: The installer fails with
Can't locate DBI.pm in @INC. It seems the internal cPanel Perl environment is corrupted or incomplete. -
Filesystem Lock: I am unable to perform a clean re-installation because the directory
/usr/local/cpanelcontains immutable files. Commands likerm -rfandchattr -R -ireturn "Operation not permitted," even as root. -
Environment: The server has a broken RAID 1 configuration (Disk A: AlmaLinux 8, Disk B: AlmaLinux 9). I am installing on Disk B.
I have already installed
MariaDB-develandperl-DBD-MySQLat the system level, but the installer still fails.I have enabled the Hetzner Rescue System if needed, or I can provide SSH credentials for the current AlmaLinux 9 environment. Please advise on how to proceed with a forced clean installation."
0 -
-
You shouldn't be installing *anything* before a cPanel installation. It should be a clean installation and then you would immediately install cPanel after that.
You can create a ticket through https://docs.cpanel.net/whm/support/create-support-ticket/#open-a-technical-support-ticket since you don't have access to a server at the moment.
0 -
I want to clarify that I was indeed performing a clean installation on a fresh AlmaLinux 9 environment. The files and errors currently on the system are remnants of the cPanel installer itself after it failed during the initial run. It is not pre-installed software.
I currently have full SSH access, but I am unable to clear the corrupted files (immutable attributes) or fix the Perl DBI dependency errors created by the failed attempt. I am trying to open a technical ticket, but the system identifies my license as "Partner Supported" by Hetzner.
Since this is a specific internal installer failure that Hetzner cannot resolve, could you please advise on how to bypass the partner check so your technical team can assist with this "stuck" installation?
0 -
In general, we don't clean "stuck" installations like this - we would just recommend to build another system and perform a fresh install.
Since your license is purchased through Hetzner they should be the ones offering you support, and then escalating the issue to us if needed.
0
Please sign in to leave a comment.
Comments
7 comments