The CentOS Foundation has just announced that instead of basing future versions of CentOS off of RHEL, the tried and true enterprise operating system, they are shifting their focus to “Stream” which (as I understand it) will become the upstream development branch for RHEL.
So rather than being a community version of an already-establushed RHEL OS, it’s now going to be the development version prior to RHEL’s stable release.
We need to see how the community responds and whether IBM decides to double down on this position and how they manage CentOS 8 Stream going forward. ApisCP works with C8S out of the box, in fact from an E2E this morning, builds just fine. I moved the demo server over to C8S because the best way to test if a parachute works is jump with it.
CentOS will be to RHEL as Rawhide is to Fedora. That’s not exactly a bad thing. RHEL is the de facto OS for banks, hospitals, government institutions, etc. They have funding to staff competent engineers and their reputation affirms that capacity, but these organizations that make up Red Hat’s core hate change. For hosting, we’re dealing with a lot of emerging technology that doesn’t move at the same pace as RHEL releases. I reported a bug in tmpfiles that took a year to move a fairly simple patch upstream. e is the secret sauce in tmpfiles that makes it web scale…
More recently there’s a nasty bug with systemd-resolved that caused it to fail on startup due to intrinsic changes in systemd’s security policies. It took 5 months to incorporate it into RHEL after reporting; 8 months after reporting it’s still MIA in CentOS. I had to roll out a hotfix ahead of CentOS for this. Having quicker releases with the engineering talent of Red Hat allows these fixes and features to be woven into a stable OS faster, which in turn could allow Red Hat to play catch up with Canonical for emerging tech, who still have to prove they can manage 10 year lifecycles at 1/30th the revenue of Red Hat.
Could IBM sunset CentOS? No, I think it’s too valuable to them. They need a springboard for RHEL. As a peer to RHEL, I don’t think CentOS delivers an appropriate value. Moreover, lest we forget that Red Hat is still comprised of GPL/MIT licensed software and Gregory Kurtzer, the original founder of CentOS, looks to be firing up his engine again.
Let’s play the other angle that it is a shitshow. In that case I’ll shunt development and assess alternative distributions over the Summer to position ApisCP 4 on a new distro. Knowing what uncertainty lies ahead, it’s always good not to code yourself into a corner through good architecture. Wouldn’t want to end up the next cPanel whose future has grown more improbable upon this announcement.
I use CentOS everywhere, just actually prepared new Ansible scripts and customized some packages so I can deploy new nodes on CentOS 8…
It’s time to ditch it. My servers are not a testing playground, I need stability (both OS and common packages), which CentOS provided.
Looking at all the solutions on the table, the best one seems to be to just use Oracle Linux Oracle Linux: A better alternative to CentOS
which is what I shall test soon. It is basically what CentOS is (or soon used to be), but with additional support and funding from another gigantic company.
It is Oracle, but I mean… seems usable. As long as the system doesn’t display a prompt E$$$$$ - insert a coin to continue, I’m happy. In theory, all my customized packages, configuration and automation should work without any modifications necessary.
If that doesn’t work out, my plan is to switch to the (open)SUSE ecosystem instead.
Ps: I hate Ubuntu and dislike Debian, so that’s not a viable migration path.
And now CloudLinux decides to fork and continue CentOS as a CloudLinux Community driven version from Red Hat, also with a script to migrate easily from CentOS 8. They plan to release this next year, with End of Life same as CentOS.
It didn’t take long to repurpose it into a garbage chute. I’m seeing devel packages in PowerTools desynced from their counterparts from Base. Red Hat’s got a lot of work to do with Stream if they want to preserve trust they’ve built over the years.
Ansible 2.9.16 isn’t available yet either, which breaks with their latest systemd feed.
Fortunately CentOS 7 is maintained until 2024. I started a thread for this to share my thoughts and plans and CentOS 8 news as it comes with my Centmin Mod members at https://community.centminmod.com/centos8stream/ and from the info there we know
Redhat has stated it will not block it’s source rpms or others from rebuilding from their sources so Rocky Linux and other rebuild projects can go ahead
Cloud Linux Q1 2021
CentOS 8 Stream is not a replacement for CentOS 8. But CentOS 8 Stream isn’t as buggy as folks make it out to be as it’s already tested twice, once internally and once via public t_functional test suite before stuff lands in CentOS 8 Stream and CentOS 8 Stream is only a rolling release for major versions and not minor versions.
Cloud Linux and Rocky Linux are reaching out to each other to collaborate hopefully too.
Guess I am fortunate that Centmin Mod’s CentOS 8 compatibility developments were still a work in progress while waiting on EPEL 8 to reach package parity with EPEL 7 https://community.centminmod.com/threads/centmin-mod-centos-8-compatibility-worklog.18372/. My thinking was what would be worse than release CentOS 8 support only for end users to find EPEL 8 packages that end users rely on not being available. But didn’t think there would be a worse case, CentOS 8 EOL in 12 months at end of 2021 !
Centmin Mod did at once stage work/develop/test for Oracle Linux 6 compatibility ages ago, but with CentOS 7 release, focus was back on CentOS 7.x.
cPanel and other web hosting panels, probably would go Cloud Linux route and rest of community would support Rocky Linux
But ultimately what broke is end users trust in CentOS. So guess that will drive folks to these 3 contenders. After all what good is there in control panels developing and support an OS where the actual end users and even distro package developers don’t have the demand and/or trust in?
I always feel like a caveman while using it The installer is weird, etc.
With that said, I just created an Oracle Linux 8 template image for my hosting platform. Zero issues, kickstart files used for CentOS 8 work with OL8 basically without any changes required.
I am surprised just how compatible these OSes really are.
The interesting part is that the OS itself comes with two kernels:
An “Unbreakable kernel”, which at the moment is “5.4.17-2011.7.4.el8uek.x86_64”. It’s supposed to be maintained by Oracle.
Thanks for replying and providing us with your takes on all of this, @nem and @eva2000. Super interesting to hear your perspectives on this and the potential impacts to your endeavors.
Seems only time will tell what direction the community will take with the forks already underway and see where conformity drifts to. Glad to hear both of your projects are well positioned to adapt and overcome the changes headed your way.
Right now we see a downfall of CentOS, the Linux with a very long term support. Because many technicians are unhappy, every large company wants a piece of this pie, to increase their potential sales on commercial products. It’s hard to believe someone at this point, because rarely a big company engages in something free, on a long term, out of passion for servers and desktops in general. Usually corporations want more money.
My first priority would be always to my existing Centmin Mod users and most are CentOS RPM leaning folks, so I’d definitely need a path forward for them after 2024 for a CentOS 8 alternative which is RPM/YUM/DNF based at least. Centmin Mod Ubuntu/Debian probably more likely that before but long way off for now. All development, testing and maintenance is done solely by me in my spare time for Centmin Mod. I still have my every day/paid client work I need to juggle And well all my clients are RPM/YUM based so that is a big factor.
cPanel’s announced path Cent OS 8 End-of-Life Announcement | cPanel Blog seems close to what I am thinking right now with the exception on CentOS 8 (I don’t see a point in investing in that if I haven’t really completed and released CentOS 8 support when it ends next year anyway). Not sure about Oracle Linux 8, it’s probably the easiest and quickest OS (relatively ease) for me to update Centmin Mod to support. But if there is no demand for it by end users and if web hosts/VPS providers don’t provide an easy out of box ISO/Image for instant deployment, then there is no point in supporting Oracle Linux 8. Web hosts probably more likely to provide Rocky Linux and Cloud Linux out of the box images I suspect.
Redhat/CentOS have also posted an update to clarify what CentOS 8 Stream will become - basically a continuous delivery stream for Redhat nightly minor version updates which end up upstream as Redhat Errata updates https://blog.centos.org/2020/12/centos-stream-is-continuous-delivery/. Yes there’s more automated testing done before it actually lands in CentOS 8 Stream.