D&C GLug - Home Page

[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]

Re: [LUG] It's CentOS, Jim, but not as we know it...

 

On 10/12/2020 11:23, Simon Avery wrote:
    This popped up on my feed today, and it doesn't seem like good news.


It's not. I've spent the past 48 hours focusing on this heavily as my employer is heavily invested in Centos. As a  project I've spent 18 months migrating a large number of Centos 6 production machines to Centos 8 based on the EOL for the latter being given as 2029, which is in line with RHEL 8's EOL (which as Centos is based off RHEL is logical)  (Note: Centos 7 is unaffected, and it's EOL is unchanged at 2024)

All that work is now very much at risk.

This week, that EOL for Centos 8 was changed to December 2021, shaving 8 years off support. That's huge, especially for businesses who form the vast majority of Centos installs.

Redhat blamed the Centos community for that "miscommunication" which has upset them a lot, since Redhat failed to tell or consult even the Centos board about that. (Who turn out to have zero power, with Redhat having full veto over any Centos decision)

Instead of the Centos 8 Linux, we now get Centos 8 Stream which was supposed to run in parallel. This is significantly different in that it is now the /upstream/ of RHEL, rather than /downstream/ of it.

So, Centos 8 Stream becomes the Sandpit / beta version of RHEL. That means less stability and more rapidly changing packages. Updates are good, but most companies will value stability over freshness, which was always Centos' strong suit. That's no longer true.

So why?

IBM bought Redhat for a lot of money. I think they're pushing down on the top to get more profit from the company but not really caring how Redhat find that money. Redhat stress that they made this decision, not IBM. Whether that's true or not I don't know, but having spoken to several Redhat employees, there does seem to be a feeling in Redhat of "Why should we help support Centos, how does it fit in our model?" That feels true, and was something that some people pondered when Redhat bought Centos some years ago.

Some people will say "If you want RHEL, pay Redhat" - and that's fine. But it simply is not affordable for many individuals or businesses. Many people also want the product but don't need or want to pay for the support.

The Centos-devel list and Freenode IRC channel have been... lively. And sad. Mostly sad, I think. The community feels betrayed, remember that many thousands of high skilled hours go into Centos without payment. Reading the pain from some of these people is hard. Things like "I've contributed to Centos for 17 years, it's basically my life's work. I'm heartbroken". Others who've spent hundreds of hours working towards improvements that are now completely binned, without warning. Wasted. Several have already left the project and a large number of others are still processing the news. It's awful.

Centos will definitely be a lesser product because of this. And likely, so will RHEL.

So what next?

Redhat won't reverse this decision, that much is clear. Some options for anyone pondering are:

- Change to Centos 8 Stream and pray like hell it's not as unstable as it might become.
- Pay for RHEL. (Ethically I now have a problem with this)
- Move to an alternative OS

     1. Centos like:
         1. CERN linux (proven)
         2. Scientific Linux  (Poss abandoned)
         3. Oracle Linux (yuck) (Ethically I have a problem with this)
         4. Rocky Linux (Emerging)
         5. (Non commercial users might want to explore Rhel Free)
     2. Alternative
         1. Debian Stable
         2. Ubuntu Server


If you're using centos, the bottom line is - don't do anything now.

There are 12 months of support for Centos 8 left, and things are very much up in the air. New distros will emerge, many will fail. Some will get the support of those now wholly disenfranchised by Redhat, and hopefully, one or two will flourish.

The writing was on the wall for this not when IBM bought RedHat but when RedHat borged CentOS a while back and brought it fully under their control.

What a sting for you though - I know others in the same situation who basically had the CentOS roadmap pinned to the office noticeboard because it was that critical a part of planning ahead... for years out. I've got more than a few CentOS instances to look after myself.

Too many people - unwisely in my opinion - have been taking their freebie enterprise-grade RHEL clone completely for granted. Let's call them the free-as-in-beer enthusiasts, rather than the free-as-in-freedom crowd. Let's face it, CentOS was always a bit too good to be true from the boring business side of things: literally full RHEL but with no monetary cost associated. Technically speaking, CentOS is just another conservative Linux distro and there are plenty more of them. And this showed up a lot over the projects' history - always chronically underfunded and struggling for resources, vampiring it's hard working contributors efforts and lagging behind upstream for years and years. Despite having by far the biggest deployment footprint of any Linux! And despite most of these users being 'proper', commercial users as well, thousands and thousands of them.

Yet they all did nothing for nearly two decades while stuffing CentOS into every mission critical bit of infrastructure they had. If some of the tens of thousands of companies literally running their businesses off CentOS had contributed back a little then it would have been able to stay healthy, independent and RedHat wouldn't have re-absorbed it and then shot it in the head. CentOS was always used by 99% leeches - unique in the Linux distro world.

The good news is that CentOS - which as you may have guessed, I don't like very much - will soon largely disappear from the internet and my life, good riddance. It reminds me of all the nasty enterprisey UNIX distros from the past that I used to love so much, and not in a good way. Even better news is that SO MANY PEOPLE rely on it that this will be basically fixed in a mad scramble so sysadmins won't have to do that much anyway.

CentOS 8 is supported for 12 months and 7 is still supported out to 2024 as is. By the time CentOS 8 is getting ready to EOL next year Rocky Linux or whatever other community led projects to basically redo CentOS _again_ will have matured to the point they'll probably just be a drop-in. And if not, well, there's plenty more Linux variants out there and most of them are better anyway ¯\_(ツ)_/¯

Let's hope the freeloading CentOS v2 commercial users who all just got bitten learn their lesson this time around - the code itself may be free as in freedom but nothing else is. Chip in with some pennies to support the operating system project that you literally depend on for your business or else.


* Better add a specific disclaimer that this isn't aimed at you Simon, or your employer (I imagine you're both CentOS "users" and community contributors rather than leeches).

I'd also like to add that everyone is looking in the wrong place anyway - the future is containerized. RedHat's Atomic/Silverblue is the way things will go, not crappy old monolithic legacy installs. That and the cloud (including "on-premises" clouds) will shortly blow away what's left of what people think of as normal Linux servers these days.

--
The Mailing List for the Devon & Cornwall LUG
https://mailman.dcglug.org.uk/listinfo/list
FAQ: http://www.dcglug.org.uk/listfaq