Resolved Installing PHP 8.2 Gives Error

[scsa20]

Username:

TITLE

Installing PHP 8.2 Gives Error

PRODUCT, VERSION, OPERATING SYSTEM, ARCHITECTURE

Plesk Obsidian v18.0.49_build1800221221.11 os_CentOS 7
CentOS Linux 7.9.2009 (Core)

PROBLEM DESCRIPTION

When installing PHP 8.2 it gives an error of "Unable to create yum cache for '*epel' OS repository."

STEPS TO REPRODUCE

1. Log into Plesk and go to Server Management > Tools & Settings
2. Click on "Updates" under Plesk
3. Click on "Add/Remove Components"
4. Plus down on Web hosting > PHP interpreter versions
5. Mark PHP 8.2 for install
6. Click Continue

ACTUAL RESULT

Will error with "Unable to create yum cache for '*epel' OS repository."

EXPECTED RESULT

For it to just install

ANY ADDITIONAL INFORMATION

Troubleshooted this for a bit and it seems like this error happens when there's already cached data for epel. You can see the tail log of when trying to run the updater to install PHP 8.2 the first time here.

After issuing yum clean all to clear to repo cache and run through the installation, it runs through fine as seen in this tail log here.

I would expect the existing cache shouldn't be an issue but now I'm thinking the installer should at least clear the cache prior to actually processing through everything else for something like this to not happen.

YOUR EXPECTATIONS FROM PLESK SERVICE TEAM

Confirm bug

 

[Peter Debik]

Thank you for your input. I have forwarded it to developers for further checks as [PPS-13779]. However, it may be a question of philosophy whether the software should be allowed to clear the cache or whether that would need to be done by a system administrator before installing new packages. We'll see what developers think about it.

 

[Peter Debik]

Developers have looked into this case but respond that the behavior is expected and the operation should not be included in the installation flow. At the first glance, it might be safe but there are various possible issues that might happen in the future. For instance, some packages might stop updating completely (e.g. MariaDB). We believe that most OS-related operations have to be handled by the person who manages the server. Also, it is not an uncommon situation when the execution of the command yum clean all is not enough. Sometimes, it is necessary to remove the cache manually (rm -rf /var/cache/yum).

Moreover, you may see that even RedHat themselves used to integrate this recommendation:
https://access.redhat.com/discussions/3195102
(It's shown when the yum clean all is executed on a system.)

Bottom line: This cannot be considered a product issue.

 

[scsa20]

Alright, thanks for the update on that.