• If you are still using CentOS 7.9, it's time to convert to Alma 8 with the free centos2alma tool by Plesk or Plesk Migrator. Please let us know your experiences or concerns in this thread:
    CentOS2Alma discussion

Question CentOS2Alma conversion: Your expectations and experience?

Peter Debik

Community Manager until 3/2024
Plesk Team
Server operating system version
CentOS 7.9
Plesk version and microupdate number
Obsidian
Hi Pleskians,

many of you are still using CentOS 7.9. Support for CentOS 7.9 will end June 30th, 2024. Since 2023 Plesk provides a tool to convert a CentOS 7.9 installation to an Alma 8 installation. The conversion tool does pre-upgrade checks, then converts the operating system including repositories and binaries that need to be upgraded, then also upgrades Plesk to the Alma 8 version. The script is available for free at GitHub - plesk/centos2alma: CentOS 7 to AlmaLinux 8 conversion tool .

I wonder who of you has already used it and what your experiences where? What were your expectations and have these expectations been met? Do you plan a conversion or rather migrate your CentOS 7.9 servers to another operating system using Plesk Migrator?

I'd love to discuss this here in more details. Please join the discussion with your thoughts.
 
  • Conversion inside containers (like Virtuozzo containers, Docker Containers, etc) are not supported.
Hmm... So, does it seem impossible to use this script for the DigitalOcean Plesk server?
 
Hmm... So, does it seem impossible to use this script for the DigitalOcean Plesk server?
The underlying Leapp project does not work for containers, but I am not sure about DO. I've asked the developer for details and will come back to the topic once I know more.
 
Today I did another test with two specific settings: a) On the new test server, the /tmp was not a partition, but I created it as a file, put it into /etc/fstab and mounted it as /tmp. b) Older servers might still use Courier instead of Dovecot, so I wanted to know whether this will work with a conversion, too. Again this new test went through without errors.

A day ago I tested the conversion of a server that had 250+ customer accounts and 1000+ domains on it. Conversion went through without issues. It took a total of only 45 minutes.

I am still interested in hearing more results from you guys. Let us know if you experience any issues or have questions.
 
The attempt was unsuccessful.
After fixing the first error about libwebp7 package dependencies on the first stage, I see:
Code:
# ./centos2alma
Do preparation checks...exec: "mysql": executable file not found in $PATH
Looks like mariadb is gone for some reasons...
 
Thank you for testing it. I've forwarded the post to Mikhail S for further consideration.
 
I posted the issue to github. In general, the point is that it is better to upgrade MariaDB to version 10.6 first, before running the script. After that the conversion was done. But now there are a lot of post-conversion issues. Now I'm sitting here fixing everything in order. It is especially bad that the mail has fallen off. Websites were able to raise relatively easily.
 
We are looking forward to run this conversion tool on 9 CentOS servers. Currently in the planning stages. We are going to run this tool on cloned servers first.
 
I did a few more tests yesterday and today and found two things:
1)
a) If the network connection is interrupted (e.g. by a router failure) during stage 1, simply reconnect the network an run the conversion again. it will work.
b) Stage 2 does not need a network connection at all.
c) If the network connection is interrupted during stage 3, the final stage, reconnect the network, then simply rerun "./centos2alma --finish", it will correctly complete the conversion.
--> It seems to be rather failproof even for the case that an unstable network connection occurs.
2)
When you have SELinux set to "disabled", the conversion will change the status to "permissive". This will trigger auto-relabelling on reboot. Auto-relabelling can last a very long time, depending on the number of files on the system, and it might appear that the system became unresponsive during that time (e.g. 10 to 90 minutes), but it is not. Normally, just wait for it to complete. If you want to speed up auto-relabelling, create a text file /etc/selinux/fixfiles_exclude_dirs with two lines: /var/www and /home. This will exclude those directories from auto-relabelling by SELinux and shorten the duration to around 2 to 7 minutes.
 
I can confirm that the script is looking good. At least we won‘t be able to migrate that soon due to the fact that we still have more than 100 customers using actively PHP 5.3.

Totally understand that the old PHP is out of support for a long time already, but many applications do not run on newer PHP versions and it still will take a long time to recode.
 
@Peter Debik this is a good question, that's going through my head every day I would say. We even see more and more customers migrating to us because we can still deliver PHP 5.3 support. Funny thing is, that we even have more people asking for PHP 5.3 than PHP 5.6.

Currently we have the following options, still to be evaluated:
  1. Setup a new server with CentOS 7 (or stay with the existing one), install Plesk and use it without support. As most probably we won't receive any Plesk updates after OS is EOL, this is a pretty expensive method.
  2. Setup a new server with Debian 11 or 12 and install Plesk & old PHP versions (need to verify if the PHP 5.3 is available through Autoinstaller there)
  3. Setup a new server with supported Debian & CentOS without using Plesk to host the old websites
  4. Try to develop a PHP installation routine for Alma 8
What we definitely plan to do is to split into servers for "old" customers and "actual ones"... But, that being said, new customers means PHP 7.0 and PHP 7.2.
From far more than 1.000 domains we have only a hand-full customers using PHP 8.

Honestly I do not think we are the only customer using PHP 5.3. CentOS 7 is still a great OS, our servers run without any issues and we support up to the latest versions of all (web applications).
 
I'm looking to run this, however I'm running Plesk with Centos7.9 inside a Proxmox LXC container. It doesn't use Virtuozzo anymore and now uses plain LXC containers.
Will this script work in that scenario?
 
@galaxy No, conversions inside containers are not supported. Instead, please use the free Plesk Migrator extension to migrate the content to a new server.
 
@galaxy No, conversions inside containers are not supported. Instead, please use the free Plesk Migrator extension to migrate the content to a new server.
I can't do the migration. I've customized the system. I learned years ago that a migration isn't an option. It must be an in-place upgrade or I can't do it.

I can convert the LXC container to a VM. If I do that, can the VM be upgraded? I'd then convert the VM back to a container.
 
Back
Top