Convert a CentOS 7 server with Plesk to AlmaLinux 8

CentOS 7 to AlmaLinux 8 conversion tool

Introduction

This script is the official tool for converting a CentOS 7 server with Plesk to AlmaLinux 8. It uses the AlmaLinux ELevate tool, which is based on the leapp modernization framework. The script includes additional repository and configuration support provided by Plesk.

Preparation

To avoid downtime and data loss, make sure you have read and understood the following information before using the script:

1. Back up all your databases and have the means to restore them. The script uses standard MariaDB and PostgreSQL tools to upgrade the databases, but this does not guarantee that the process will be free of issues.
2. Ensure that you have a way to restart the server without a direct SSH connection. The conversion process may get stuck once the server boots into the temporary OS distribution that does not start any network interfaces. You can use a serial port connection to the server to monitor the status of the conversion process in real time, and to reboot the server if necessary.
3. We strongly recommend that you create a snapshot you can use as a recovery point in case the conversion process fails.
4. Read the Known issues section below for the list of known issues.

Timing

The conversion process should run between 30 and 60 minutes. Plesk services, hosted websites, and emails will be unavailable during the entirety of the conversion process. The conversion process itself consists of three stages:

• Preparation, which takes between 15 and 25 minutes.
• Conversion, which takes between 15 and 30 minutes. During this stage, the server will not be available remotely. You can monitor the progress via a serial port console.
• Finalization, which takes between 5 and 10 minutes.

Known issues

Blockers

Do not use the script if any of the following is true:

• You are running an OS other than CentOS 7.9. The script was not tested on other Red Hat Enterprise Linux 7-based distributions. The conversion process may have unexpected results if started on a server not running CentOS 7.9. So we add checks to avoid any actions on such kinds of servers.
• Plesk version is more than five releases behind the latest version. The script is only compatible with the most recent versions of Plesk. It will prevent conversion if Plesk version is outdated.
• PHP 5.5 and earlier are not supported in AlmaLinux 8, and will not receive any updates after the conversion. These PHP versions are deprecated and may have security vulnerabilities. So we force to remove this versions before the conversion.
• Conversion inside containers (like Virtuozzo containers, Docker Containers, etc) are not supported.
• More than one kernel named interfaces (like ethX) are not supported. Stability of such names are not guaranteed, so leapp prevent the conversion in such cases.
• No ELS repositories on the server. The script won't be able to install the necessary packages at the final stage due to conflicts between AlmaLinux 8 repositories and ELS repositories. However, the script can handle ELS provided by the Plesk TuxCare ELS extension.

Requirements

• Plesk version is no older than five releases back from the latest version
• CentOS 7.9 or later
• GRUB 2 is installed
• Minimum of 5 GB of free disk space
• Minimum of 1 GB of RAM
• The Linux kernel package must be signed by Red Hat. This is typically the case if you have installed the kernel from the official CentOS repositories

Using the script

To retrieve the latest available version of the tool, please navigate to the "Releases" section. Once there, locate the most recent version of the tool and download the zip archive. The zip archive will contain the centos2alma tool binary.

To prepare the latest version of the tool for use from a command line, please run the following commands:

> wget https://github.com/plesk/centos2alma/releases/download/v1.4.4/centos2alma-1.4.4.zip
> unzip centos2alma-1.4.4.zip
> chmod 755 centos2alma

To monitor the conversion process, we recommend using the 'screen' utility to run the script in the background. To do so, run the following command:

> screen -S centos2alma
> ./centos2alma

If you lose your SSH connection to the server, you can reconnect to the screen session by running the following command:

> screen -r centos2alma

You can also call centos2alma in the background:

> ./centos2alma &

And monitor its status with the '--status' or '--monitor' flags:

> ./centos2alma --status
> ./centos2alma --monitor
... live monitor session ...

This will start the conversion process. During the process, Plesk services will stop, and hosted websites will not be accessible. At the end of the preparation stage, the server will reboot. Next, a temporary OS distribution will be used to convert your CentOS 7 system to AlmaLinux 8. This process will take approximately 20 minutes. Once completed, the server will reboot once more. The centos2alma script will then perform the final stages of reconfiguring and restoring Plesk-related services, configurations, and databases. This will take some time, depending on the number of hosted websites. Once the process is complete, the centos2alma script will reboot the server one last time. After that, Plesk should return to normal operation. On the next SSH login, you will be greeted with the following message:

===============================================================================
Message from the Plesk centos2alma tool:
The server has been converted to AlmaLinux 8.
You can remove this message from the /etc/motd file.
===============================================================================

Conversion stage options

The conversion process consists of two stage options: "start", and "finish". To run stages individually, use the "--start", and "--finish" flags, or the "-s" flag with name of the stage you want to run.

1. The "start" stage installs and configures ELevate, disables Plesk services and runs ELevate. It then stops Plesk services and reboots the server.
2. The "finish" stage must be called on the first boot of AlmaLinux 8. You can rerun this stage if something goes wrong during the first boot to ensure that the problem is fixed and Plesk is ready to use.

During each phase a conversion plan consisting of stages, which in turn consist of actions, is executed. You can see the general stages in the --help output and the detailed plan in the --show-plan output.

Other arguments

Logs

If something goes wrong, read the logs to identify the problem. You can also read the logs to check the status of the finish stage during the first boot. The centos2alma writes its log to the '/var/log/plesk/centos2alma.log' file, as well as to stdout. The ELevate writes its log to the '/var/log/leapp/leapp-upgrade.log' file. Reports can be found in the '/var/log/leapp/leapp-report.txt' and the '/var/log/leapp/leapp-report.json' files.

Revert

If the script fails during the the "start" stage before the reboot, you can use the centos2alma script with the '-r' or '--revert' flags to restore Plesk to normal operation. The centos2alma will undo some of the changes it made and restart Plesk services. Once you have resolved the root cause of the failure, you can attempt the conversion again. Note:

• You cannot use revert to undo the changes after the first reboot triggered by centos2alma.
• Revert does not remove Leapp or packages installed by Leapp. Neither does it free persistent storage disk space reserved by Leapp.

Check the status of the conversion process and monitor its progress

To check the status of the conversion process, use the '--status' flag. You can see the current stage of the conversion process, the elapsed time, and the estimated time until finish.

> ./centos2alma --status

To monitor the progress of the conversion process in real time, The conversion process can be monitored in real time using the '--monitor' flag.

> ./centos2alma --monitor
( stage 3 / action re-installing plesk components  ) 02:26 / 06:18

Special cases

Postgresql database before version 10 is installed

By default, the tool does not allow conversion when a PostgreSQL database version prior to 10 is installed. This restriction is in place to warn about the potential loss of data during the PostgreSQL upgrade that will be performed as part of the conversion process.

In such cases, you have two options:

1. Upgrade PostgreSQL to version 10 manually before initiating the conversion.
2. Create a complete backup of the database and force the conversion using the '--upgrade-postgres' flag.

Adjusting Leapp Overlay Size

The default Leapp overlay size is 2048 MB. In the centos2alma tool, this is increased to 4096 MB. However, this may still be insufficient for handling all the required upgrade packages, leading to a "Disk Requirements:" error from Leapp. To prevent this issue, you can increase the overlay size using the '--leapp-overlay-size' flag. For example, to set the overlay size to 8192 MB, execute the following command:

> ./centos2alma --leapp-overlay-size 8192

Perl modules installed by CPAN

During the conversion process, if the tool detects Perl modules that were installed via CPAN and cannot determine their corresponding RPM packages, it will fail with a warning. This restriction is in place because such modules will not be available after the conversion. This issue often arises because CPAN-installed modules are specifically built for a particular version of Perl. Consequently, when Perl is updated during the conversion process, these module libraries may encounter errors related to undefined symbols.

To prevent this issue, there are steps you can check for RPM package analogues for the modules, remove the CPAN modules, and then reinstall them after the conversion is complete. It is recommended to use RPM packages for reinstallation, although reinstalling the modules via CPAN is also an option.

The centos2alma tool includes a list of RPM mappings for certain modules and can automatically reinstall them. The warning will only raise for modules that mapping to rpm package is unknown by the tool.

If you are confident that you no longer require the modules installed via CPAN, you can forcefully remove them by running the tool with the '--remove-unknown-perl-modules' flag.

Issue handling

Leapp unable to handle packages

Leapp may not be able to handle certain installed packages, especially those installed from custom repositories. In this case, the centos2alma will fail while running leapp preupgrade or leapp upgrade. The easiest way to fix this issue is to remove the package(s), and then reinstall them once the conversion is complete.

Leapp cannot choose which package to install

This issue may occur if unsupported repositories are used. For instance, if an unexpected EPEL repository is enabled on the server, the cloudlinux7to8 process will fail when running leapp preupgrade or leapp upgrade. This is due to conflicting packages between the el8 and el7 repositories. The simplest way to resolve this issue is to switch to the standard repositories.

Temporary OS distribution hangs

This issue may occur, for example, if there is a custom python installation on the server. The conversion process will fail while upgrading the temporary OS distribution, and the temporary OS will hang with no notification. To identify the issue, connect to the server using a serial port console and check the status of the conversion process. To fix the issue, reboot the server. Note that an unfinished installation process may result in missing packages and other issues.

centos2alma finish fails on the first boot

If something goes wrong during the finish stage, you will be informed on the next SSH login with this message:

===============================================================================
Message from Plesk centos2alma tool:
Something went wrong during the final stage of CentOS 7 to AlmaLinux 8 conversion
See the /var/log/plesk/centos2alma.log file for more information.
You can remove this message from the /etc/motd file.
===============================================================================

You can read the centos2alma log to troubleshoot the issue. If the centos2alma finish stage fails for any reason, once you have resolved the root cause of the failure, you can retry by running:

> ./centos2alma --resume

Send feedback

If you got any error, please create an issue on github. To do generate feedback archive by calling the tool with '-f' or '--prepare-feedback' flags.

./centos2alma --prepare-feedback

Describe your problem and attach the feedback archive to the issue.