Question Rocky 9 support

[QWeb Ric]

Server operating system version

Rocky 9

Plesk version and microupdate number

18.0.60

We have 3 CentOS 7.9 servers to migrate for separate clients, and as we're already running a Rocky 8 server for another client with Plesk installed to that just fine, I perhaps ignorantly just assumed Plesk would also run on Rocky 9 without problems.

Having spent 2 days building the first replacement server, with Rocky 9, I've just now attempted the Plesk install and found that it errors out:

RROR: You are trying to run Plesk Installer on an unsupported OS.
Your OS is Rocky 9 x86_64

And indeed, as per Software Requirements for Plesk Obsidian , Rocky 8 is listed as supported but not 9.

I'm now aware that there are older threads regarding this ( Question - Rocky Linux 9 Compatibility? ) and a feature request at Plesk Support for Rocky Linux 9.0 but this request has since been closed and is no longer accepting votes.

With CentOS 7.9 hitting EOL in just a couple of months, and presumably many people now looking to migrate, it seems baffling to me that there wouldn't be a bigger push at this point to support Rocky 9! 8 has an EOL of 2029 so isn't as desirable as 9 with its 2032 EOL, and I can see AlmaLinux 9 is supported by Plesk already. As far as I understand it, Alma and Rocky are both near identical given they're binary compatible with RHEL so surely supporting Rocky 9 should be a breeze if Alma 9 already works?

I'm going to have to look at the feasibility of building Alma servers instead I suppose, but this will take further liaising with each client to get sign-off and take us even closed to the CentOS EOL deadline so far from ideal.

I'm crossing all fingers that, despite the closing of the above feature request, Rocky 9 support is actually about to land???

 

[PAM Baduijn]

I am using AlmaLinux 9.3 (Shamrock Pampas Cat) for 3 months now without any problems!
The limitations described i don't use anyway, so considering if you do not use this extension....

https://support.plesk.com/hc/en-us/articles/12376955168919-Does-Plesk-support-RHEL9-AlmaLinux-9

 

[AYamshanov]

@QWeb Ric, you have correctly identified the current state of supported OSes and available options,

• You can upgrade CentOS7 servers to Alma8. For me, it sounds like more preferable option because in the future it could unlock the upgrading path to AlmaLinux9 (EOL will be 2032).
• You can upgrade CentOS7 servers to Rocky8 (but without ability to upgrade the servers to Rocky9 since Plesk does not have plans to support Rocky Linux 9 in the nearest future), in this case EOL will be 2029. At least, this will provide additional time to talk with end-customers and weigh all possible options for further steps.

 

[QWeb Ric]

@QWeb Ric, you have correctly identified the current state of supported OSes and available options,

• You can upgrade CentOS7 servers to Alma8. For me, it sounds like more preferable option because in the future it could unlock the upgrading path to AlmaLinux9 (EOL will be 2032).
• You can upgrade CentOS7 servers to Rocky8 (but without ability to upgrade the servers to Rocky9 since Plesk does not have plans to support Rocky Linux 9 in the nearest future), in this case EOL will be 2029. At least, this will provide additional time to talk with end-customers and weigh all possible options for further steps.

Fair point on upgrading in steps, but our servers are all cloud instances so unfortunately the CentOS to Alma upgrade script isn't an option, which is a shame as that does sound like a brilliant solution.

It seems the sensible approach is going to be for me to start again with the server build, going with Alma 9 this time as it's already supported and given Plesk has gone to the effort of making the aforementioned upgrade script, presumably they plan to offer long term Alma support.

Utterly mind boggling though, and frustrating, that there wouldn't by now just be proper support for Rocky 9 by a system that already has Alma 9 and RHEL 9 support. Extending to Rocky 9 surely isn't as costly as we're being told? We're all paying for Plesk licenses and they don't exactly come cheap, so it's not liike those of us on CentOS 7.9 haven't essentially been paying for Rocky support to be worked on over the last couple of years.

I appreciate Plesk does an awful lot that would need tweaking and testing for actually new distributions, but Rocky and Alma are as far as I understand it pretty much identical from a configuration and software support perspective, given they're both RHEL downstreams, so what works in one should theoretically work the same in the other?

 

[Bitpalast]

so what works in one should theoretically work the same in the other?

Yes, in theory. In practical application I've been through dozens of "surprises" even with the conversion from CentOS to Alma although all started out from seemingly perfect "default" systems. I bet there will be hundreds of deviations between Rocky and Alma. You have to figure that every Plesk change/update must be tested against all officially supported operating systems. Can you really guarantee that Rocky and Alma are binary-identical? One small glitch and it could lead to complete outages of a server. So when Plesk decides to officially support a new operating system branch, they'll also need to provide full tech support, must have a complete set of test servers, test routines etc. Plus, when you argue that Rocky is identical to Alma, then why don't you just choose Alma?

Rocky is used by a very low number of Plesk users. It is understandable that by economic aspects it is much better to focus on one of the two (Alma and Rocky) to save resources and provide best industry support for one of the two than to provide mediocre support for both.

 

[QWeb Ric]

Yes, in theory. In practical application I've been through dozens of "surprises" even with the conversion from CentOS to Alma although all started out from seemingly perfect "default" systems. I bet there will be hundreds of deviations between Rocky and Alma. You have to figure that every Plesk change/update must be tested against all officially supported operating systems. Can you really guarantee that Rocky and Alma are binary-identical? One small glitch and it could lead to complete outages of a server. So when Plesk decides to officially support a new operating system branch, they'll also need to provide full tech support, must have a complete set of test servers, test routines etc. Plus, when you argue that Rocky is identical to Alma, then why don't you just choose Alma?

Rocky is used by a very low number of Plesk users. It is understandable that by economic aspects it is much better to focus on one of the two (Alma and Rocky) to save resources and provide best industry support for one of the two than to provide mediocre support for both.

I'm not on the Rocky / Alma / CentOS / RHEL dev teams so can't guarantee anything, granted. But Rocky was founded by Gregory Kurtzer, one of the original CentOS founders, so I'd expect it to be as close, if not closer, to a binary compatible RHEL than Alma, given both distributions make that claim.

As for my preference for Rocky over Alma - basically as above, I liked that Rocky was founded by Gregory since it's CentOS I'm most familiar with. But also, the last I checked Alma was funded by a $1m yearly donation where Rocky was founded with something like a $30m initial donation so financially should be a more stable development. It's basically long-term support without the cost of a RHEL license that's the most appealing to me, for a server O/S.

As above, Plesk does seem to be going with long term support for Alma so this is where I'm going to transition to, but after the sudden demise of CentOS right after they'd just launched CentOS 8 with a 10 year EOL promise, I don't feel quite as stable spinning up an Alma server as I did spinning up a Rocky build.