The Centos move to Centos Stream

I’d love to hear your take on Centos’s move to Stream. My gut is that it is bad for me. I chose Centos for my VPS because I wanted its stability and its promised 10 years of security updates. But Centos Stream, as I understand it, gets maybe 6 years of support and packages do not go through the same level of testing that the normal Centos releases go through.

1 Like

Good question, I was wondering the same. I was very excited about CentOS Stream when it came out and tried it for some months in a VM. Frankly found it a little rough around the edges and not very stable, unlike CentOS proper, which basically was RHEL without branding. I think CentOS will be a real lost. CentOS stream has value, but which enterprise would be wanting to test software before it enters RHEL? I thought the whole point of CentOS was that enterprises who were very low on funds could use it, then step up to RHEL when/if they were able…

1 Like

I just caught the Ars Technica piece on this a few minutes ago. My gut reaction is that this is a bad decision on Redhat’s part and will only breed ill will in the community. I remember using CentOS as a trainer for Redhat in college.

According to another podcast, there is a rumor that Redhat will offer free versions of rhel for personal use.

But I have always preferred Debian/Ubuntu so when Centos 7 expires, I’ll likely switch to the current Ubuntu LTS. I switched to Centos in the first place because of the 10 years of support.

They already do…sort of. They’re supposed to be for “development purposes only.”

There are at least 2 fork of rhel 8 that have been announced because of this.

and

At work we mostly use CentOS so we’re hopefull about these projects. I really like that cloudlinux is using their experience from their commercial offering to provide a fully open fork.

In their words:

Why We Are Doing It

  1. We have all the infrastructure, software and experience to do that already. We have a large staff of developers and maintainers that have a decade of experience in building an RHEL fork, starting from RHEL5 to RHEL8.
  2. We expect that this project will put us on the map, and allow people to discover our rebootless update software and Extended Lifecycle Support offering.
1 Like

I hope so otherwise… I will just stop but I think it was the wrong move and the existing developer subscription is not something I would take seriously to work with.

Corporate interests are corporate interests unfortunately that is why we have Debian and it is independent and of course there is Canonical’s Ubuntu LTS.
I think Ubuntu will be getting more and more important and is a serious contender to Red Hat.

CentOS Stream is not for enterprise. They say it in the announcement, it is a developer preview for RHEL. In other words CentOS like we know it will be dead.

2 Likes

I read the post on the CentOS website and I’m confused by a few things. The post says

“CentOS Stream, which tracks just ahead of a current RHEL release.”

What does it mean exactly to “track just ahead”? I see some articles from ZDnet calling it a “rolling release”. If it’s “just ahead” of RHEL then it fits somewhere between RHEL and Fedora right? But when someone says “rolling release” I think Arch Linux. Which would be even more current than Fedora.

The CentOS Stream webpage says:

“positioned as a midstream between Fedora Linux and RHEL”

So why is ZDnet lying? or just really misinformed? (Are they a reputable source even?) Their own definition of rolling release states:

“the idea is that users and developers are best served by giving them the latest updates and patches as they’re created.”

Ummm… “latest” doesn’t sound like “just ahead” of RHEL or “between Fedora and RHEL” to me. They’re calling it something that does not fit their own definition of that thing.

Anyways, what I’m guessing is that RedHat saw CentOS as redundant, since you can get the free unsupported RHEL already and CentOS was the same thing. So if you want to develop or test for RHEL you use RHEL, if you want to develop and test for what-will-soon-become-RHEL then you use CentOS, if you don’t want a RedHat developer account then you use one of the upcoming forks. Isn’t that reasonable-ish?

What does it mean exactly to “track just ahead”? I see some articles from ZDnet calling it a “rolling release”. If it’s “just ahead” of RHEL then it fits somewhere between RHEL and Fedora right? But when someone says “rolling release” I think Arch Linux. Which would be even more current than Fedora.

https://blog.centos.org/2020/12/centos-stream-is-continuous-delivery/

Might explain it a little better.

2 Likes

There are proven advantages to continuous integration and continuous development which is probably why many software projects embrace the process. I have to say though I remain unconvinced how high this can be scaled up, unfortunately.

For example, before a lot of this took hold, it used to be common practice to not used the latest release of development kit (like MS Developer Studio) for long-term projects, but rather the previous release, as new features inevitably introduced bugs that would take time to iron-out and so latest releases were less suitable for production use, as-it-were…

A couple of years ago when I started using AndroidStudio, I found that as the system was more-or-less constantly updating (Gradle, IntelliJ, Android libraries, Qemu emulator etc.) At the time I tried it, even Google-provided basic templates were not compiling; whetting one’s appetite to always install new updates to see if they’d bring-in a fix, and constantly finding that the system did not work… And I tried over and over for a period of two-three-months. Now maybe I just lacked sufficient-know-how in the complexities of Gradle and its integration with everything else, but it completely put me off trying to get anywhere beyond a very simple HelloWorld Android application. As things are changing all the time, any workarounds found and documented, say as tutorials on youtube, became quickly irrelevant, as does printed documentation apart from Google’s own online documentation, which when followed, didn’t work! That was just for a development system…

I don’t even want to get started on Windows 10 and its stupendous updates, which have to be the biggest failure for continuous integration I’ve ever witnessed. Do we really want nonsense like this in Linux?! Maybe folks who use rolling releases have more confidence in systems like this, but surely they have the know-how to take the technical risk of deploying such systems? I know I really don’t want to be doing manual systems admin every time I update the system - which on Debian is just for security updates and is generally pretty flawless.

I do understand the enthusiasm behind this, and its fair enough. I have to say though, I would be particularly sceptical of a supposedly enterprise solution constantly updating like this, and being expected to remain fully functional. Just my thoughts.

1 Like

This picture was pretty excellent from their blog (Thank you grumpey)

If i’m wrapping my head around this correctly…

Instead of CentOS forking from RHEL and RHEL occasionally forking from Fedora… CentOS Stream will occasionally fork from Fedora and RHEL will occasionally fork from CentOS.

This seems like a very good move for RHEL because it not only adds another testing layer but one with a much larger user base compared to their internal point release testing.

What I don’t know yet is why they’re removing the stable release cycle of CentOS (even to the extent of cutting expected support) when they could have both stable and rolling versions like most major distros. It’d keep all the users under their proverbial community roof instead of pushing them out to the stable alternatives being created or even out of the RHEL ecosystem.

1 Like

As I understand it, they are very limited in the number of developers for Centos (2 is the number I seem to recall) and can’t do both.

I’m wondering if the plan would be to continue only having 2 devs managing the upstream of RHEL (between Fedora forks) or if it only took 2 devs to manage CentOS stable and Red Hat will be adding more to handle Stream.

From what I know at this point, we’re* not planning additional hiring specifically for Stream — but it’s actually bigger than that, because with this change, every RHEL maintainer (numbered in the hundreds) becomes an active CentOS maintainer.

* I work for Red Hat, but on Fedora, not CentOS or RHEL directly.

4 Likes

I found out about Oracle Linux today. Linux | Oracle

Could be an alternative for lots of people instead of waiting for the CentOS forks.

There exists apparently an even older RHEL clone than CentOS that I did not hear about before these news, Springdale Linux.

Interesting. I’d trust anything coming out of Princeton.

It may introduce some complications with some vps providers.

Oracle’s Unusable Linux is not an option anyone should consider in my opinion. I mean Oracle is behind the efforts of trying to break programming. They are suing Google because they want to sue Google but in doing so they are risking the viability of APIs which means they could break software programming into a monstrosity of legal nightmares . . . I will never use something from Oracle if I can help it and in this case, I can absolutely help it.

3 Likes

They do seem a very aggressive corporation. I’m a little sad to be so invested in Java Programming, which originated with Sun, whom I trusted more, though I do think OpenJDK is definitely a step forward, as is allowing the community to forge ahead with OpenJFX, which I am currently learning for cross-platform use.

2 Likes

I agree. I should have known better than to utter the “O” word around here… I switched from VirtualBox to KVM a while back too. I haven’t de-Googled my life yet, but I’ve de-Oracled it.

2 Likes