I'd recommend anybody truly curious about this check out ep383 of the Linux Unplugged podcast. Its the only serious analysis of the shift, with people in the know, that I have seen. No, its not all roses and sunshine but the 'community' response has been so universally full of FUD and dumb assumptions.
RedHat bought the CentOS project and they did it in the service of RHEL. Its kind of amazing that something like this <<hasn't>> happened already over the last 6 years
>RedHat bought the CentOS project and they did it in the service of RHEL
It's funny, when a corporation buy's a community and then decides something without asking the community and even cut the support-time, who could imagine that people get angry?? Has RedHat the same marketing firm as Oracle?
RedHat is the major contributor to Linux ecosystem, so they are not using the free software for free. They pay back, as intended. For bad example, look at Apple or Amazon.
Sounds like a great way to stifle the innovation that tech is known for. The amount of well made free options in the tech sphere is part of why we can so efficiently start new companies.
I surely not only paid for my compilers (started coding in 1986), also do donations every time I download a distro, buy books from all open source projects that are relevant for my work.
Not everyone is as fortunate as you are, some cs students in poor countries might also need a stable server OS that is not Debian. You could argue they shouldn't but this is in the spirit of open source: non-discriminatory access regardless of the economic situation. Do not reduce it to donationaware argument. When you buy RHEL you are buying the support access and not anything else.
They do a pretty good job. Their priority is to keep things civil and authentic, so I think flames/insults/hyperpartisan politics/spam get the priority treatment.
I always found the "C" in CentOS, the "Community Enterprise OS"... awkward. I admit I haven't been following the project very closely over the years, but true community projects look different. Look at Debian, for example - internal drama and a at times gory, public decision making process. People with very different perspectives on a unfiying vision coming together to find solutions to the problems they have, and sometimes even create. It's slow, it's not as efficient as it could be - but it's a mix of democratic and meritocratic principles that is without equal, and that, I think, will stand the test of time for as long as there is such a thing as a Free Software/Open Source community to speak of.
Now CentOS's practical goal - repackaging RHEL gratis - isn't really compatible with what a true community project like Debian is about, in my view. There's no real way for the project to steer away from whatever RHEL chooses to deliver, in neither a positive or negative direction. In that sense, "CentOS Streams", as a new upstream-of-sorts for RHEL, is much more "C"-enabling that what CentOS has been before - because a community (if one will emerge, or maybe it already has) can actually take action to change their own fate, and not totally depend on whatever their Enterprise Linux Overlords decide to make happen.
A pity we're not really looking for this kind of thing where I work now, and most current users of CentOS probably weren't really in it for the "C", but more for the "entOS" part. They were looking for being able to rapidly roll out RHEL clone installations, without having to pay or care about any stupid and impeding commercial licensing restrictions at all. Well, the free lunch at that particular place is over, it seems.
That said, I'm curious to see what will take CentOS's place in that regard.
CentOS was supposed to be a community project supported by Red Hat. They appointed themselves to the 'governance board' and never held a single election, not even one seat, to represent the 'community.' Then, they made this decision behind closed doors, rather than at least have a public discussion period.
The community response is in no way FUD. They took a project, killed it, and expect you to like whatever rolling alpha they pretend is it's replacement.
RH CTO Chris Wright: "CentOS Stream isn’t a replacement for CentOS Linux; rather, it’s a natural, inevitable next step intended to fulfill the project’s goal of furthering enterprise Linux innovation."
The goal of CentOS, back when I used it, was to provide a bedrock-solid OS and package repository. CentOS machines would run for months or years without manual intervention.
That's as far from "innovation" as you can get - and that was the point. We didn't want the bleeding edge. We wanted the spine of the blade upon which to build our bleeding edge software.
The stability is, simply, not the same between them. CentOS remained on tried and tested releases for multiple years, backporting security fixes - though not features - for software. It kept the number of bugs you'd find in the software low and well known.
Let's say, Nginx. On Debian, you might be running "buster", their stable release, and running nginx 1.18-1.19. On CentOS, you'd be running on nginx 1.14. This means that you're missing features, but the versions you're running are about as bulletproof as software gets, and (were) still receiving security updates.
Newer versions are available, but usually from the developers directly; the official repos always held much older, stable versions.
When they offered "Decades of Support", they meant it (well, they used to mean it).
Can someone enlighten me, what does it mean that RedHat bought CentOS? Who owned CentOS, who got paid for it? I never did any research about those topics, but for me CentOS was like Debian: just community under a single tag which can't be bought or sold. I guess I was wrong.
In 2012-2014, CentOS had a resources issue. It was far more popular than they had resources to support. With the then upcoming switch from RHEL6 to 7, my suspicion is that they were worried about having large delays in releasing like they had at the start of the RHEL6 release.
Instead of seeing CentOS fold, Red Hat decided to hire the core members and support the project. They set up a governance structure for the project, which pretty much guaranteed a RH majority in all votes. AFAICT, there was no requirement for public voting, board seat terms, elections, or public input.
CentOS never really had the strong (public) governance model of something like Debian. RedHat bought the project the old fashioned way — they hired the people who worked on it and stacked the board.
In general Red Hat is good at infiltrating and taking control of open source projects even without hiring the main developers. The open source foundations are generally pretty vulnerable to this approach.
The problem is it never had a community to speak of, RH did aquihire it and put it on life support, now only a couple of maintainers work on CentOS, this situation is in part self-inflicted. You could blame that RH always looked at CentOS as their bastard child and never treated it as good as Fedora which has a healthy community relations. I don't have high hopes for CentOS Stream, Rocky Linux is going to be the main go-to distro for rpm from now; with no corporation standing in the way I wish them all the luck but this going to be a rocky ride.
And it's the exact opposite of what I relied upon CentOS for in the past.
In the past, CentOS was what I used when I needed a rock solid OS and software packages that could run for months, if not years, with few manual interventions.
Yeah, I myself used CentOS when I needed a 100% binary-compatible version of RHEL for free, without all the horse shit that RHEL forces you into (RHN, etc).
Oh well, I guess there's Oracle Unbreakable Linux to fill this gap until something better comes along.
Another possibility for those that are being affected by this: in the past when I had no choice but to use RHEL but didn't have an active RHN subscription, I just removed the RHN yum repos and replaced them with CentOS equivalents and everything worked out just fine.
In the event the CentOS repos get pulled, then I imagine one could do the same thing, but with Oracle's repos instead.
And when your company gets audited by Oracle and they find the repos, you’ll have a nice conversation about how you need to buy a bunch of oracle cloud bullshit or else you owe a bunch for support.
Mind! I don't mean to say that, of my own knowledge, what there is particularly dead about a doornail. I might have been inclined, myself, to regard a coffin-nail as the deadest piece of ironmongery in the trade. But the wisdom of our ancestors is in the simile; and my unhallowed hands shall not disturb it, or the Country's done for. You will therefore permit me to repeat, emphatically, that RHN was as dead as a doornail.”
You can see it like that, some fedoras (like 34) are going to be RHEL9. CentOS Stream is the beta/testfield for the next (or more then just the next) RHEL point release like from 8.3-8.4 or 8.3-8.5
As far as I can tell, it's going to be similar to Debian Testing, ie. a rolling-release distribution that will have security support. RHEL point releases will then be snapshots of that rolling release with their own security fixes.
I can see the use for such a distribution (run CI against CentOS stream to catch any potential breakage with upcoming RHEL changes) but I think the way they went about making this change is ... unfortunate.
EDIT:
To expand a bit, things may still turn out positive overall if Rocky Linux succeeds and Red Hat doesn't pull off any other silliness. There's an opportunity here for Red Hat to really open up RHEL development to interested parties via CentOS Stream, and so long as they maintain their commitment to publishing RHEL as free software and don't actively interfere with people building their own RHEL-compatible clones from the point releases, everyone wins in the end.
It's very different from Debian Testing. The closest comparison is to stable-proposed-updates - any given branch of CentOS Stream is still issuing minor updates for a single CentOS major release. The closest thing to Debian Testing would probably be Fedora, with Debian Unstable being comparable to Fedora Rawhide.
That is a very a good thing, not a downside. CentOS Stream is still a decent choice to run a server on (yes, not quite as good as the near-perfect RHEL clone that CentOS currently is), which would definitely not be the case if it was a "true" rolling release distro.
To have an Arch-like distro under the CentOS brand would not make any sense.
Right, Debian Testing is not supported as an "operating system" in any way. It's just a place for Debian developers to test their changes.
I see this misunderstanding all over the net — people think that Debian Testing is more supported than Debian Unstable, when the reverse is true. The Debian project has serious problems with public communications and expectation management.
No. openSUSE Leap is what centos was before. A community version of the enterprise distro. And unlike redhat, suse is actually moving towards converging the leap and enterprise codebases as much as they can in Leap 15.3
Yup, the only valid complaint I've seen about OpenSuse is the lower support window, "at least 36 months" for any major release.
I think most of the complainers don't realize how easy it is to upgrade OpenSuse compared to CentOS. It's like night and day.
SLES also has kernel updates with no reboot and an immutable root version. Basically / is read-only and all patches are applied to a new snapshot that is mounted at next boot.
you can't compare tumbleweed with fedora since fedora has "major releases" and a support window.
tumbleweed is basically on the edge (with transactional-update support) i.e. there is no "system upgrade"
I wouldn't expect any. It seems like Fedora will still be the upstream for major RHEL releases. This is more like a rawhide for RHEL from my perspective.
Rawhide is a bad comparison to use. CentOS Stream will never get any major fundamental changes to the system like Rawhide does on a regular basis.
It's technically rolling release compared to RHEL, but the window of acceptable changes is very small. It's not like Arch and Rawhide and Debian Testing where major OS components can be swapped out under your feet. The majority of the changes would fall into the categories of bugfix backports, backports of extremely useful/important features, and backports of support for new hardware.
This is not even remotely what the announcement is implying.
CentOS is the new upstream for RHEL minor releases. Fedora is and will continue to be the the RHEL upstream for major releases.
Lifecycle of RHEL:
Fedora
|
Enterprise Linux Next*
|
HARD FORK for next RHEL
|
Centos Stream (Pre-release phase)
|
RHEL Major release
|
CentOS Stream
|
RHEL Minor release
Those last two will loop until the RHEL full support phase (first 5 years) are over, at which point Stream will end due to there being no more minor releases. Any consumers of Stream will need to migrate to the next major version of Stream.
Stream will have effectively 6 years of support (5 + 1 year pre-release), and RHEL will be releasing major versions every 3 years.
Considering new technology and design decisions aren't incubated at all in CentOS, it is in no way shape or form possible to use it to replace Fedora.
* ELN: Fedora Rawhide with RHEL buildroot context.
Red hat is now owned by IBM. IBM needs to pull every cent of profit out of their acquisition.
Expect them to change licensing whenever possible and to use more restrictive licensing on new features to make it impossible for anyone to completely fork Redhat Enterprise in the future.
The Red Hat employees on freenode's #centos and #centos-devel have been quite adamant that IBM had nothing to do with this decision. How true that is, of course, is anyone's guess.
1. IBM being IBM and forcing RedHat to milk RHEL for all it is worth ecosystem be damned
2. RedHat deciding to do this all on their own without even requiring any pressure from IBM - because they are just as bad as IBM now
Red Hat have, contrary to the slashdot-circa-2000 meming, historically been an exemplar of open source values. They've released products that they purchased (such as RHEV) under free licenses, they do upstream delivery of their work, and they pay for a substantial chunk of the ongoing development of many key free software projects.
It was not unreasonable to assume that they would maintain CentOS.
This change has occurred now that Red Hat are IBM. I expect to see more, bad, changes. Cutting devs from free software contribution that aren't "directly important" enough, killing free software stacks like JBoss that compete with IBM offerings, and so on.
I saw some redhat people at a Linux conf and they scared me. Like they were all giddy on some cult koolaid. And they were actually strutting around wearing red hats. I maintained a healthy distance. I suspect IBM will be a perfect fit.
Alan Cox was sometimes seen with a red hat, and he's about as far from traditional blue tie as one could get. :)
Enthusiasm, and a bit of counterculture in the mix, seems like a good sign for the long-term health of the organization, IMHO.
BTW, Linux conferences at least used to be very social events, for developers who'd only been interacting online in text, to get together. Maybe people were just really happy to be there.
They always wear the hats at trade shows, it's just their promotional thing... It gives them easy recognisability, after all trade shows are all about marketing. They give them away too (but not the same quality as the ones they wear).
That giddy thing is also a tradeshow thing, some companies really pump their employees that work at these things, but they're not normally like that :)
Only if you know someone well enough you can get real info. The real communication at these tradeshows is not done in the booths but in back rooms and coffee corners. The front booths are just marketing with some low level PR staff.
Prior to the RedHat "acquisition" how was CentOS getting RHEL source? How would the "Rocky Linux" CentOS fork go about doing the same thing?
I've heard time and again that RedHat will terminate a subscriber's contract if the subscriber is caught distributing RHEL source or binaries to discourage further distribution. Detecting such redistribution seems difficult to me, but setting that aside and assuming it's true, how does a project like CentOS or Rocky Linux get RHEL sources in a sustainable manner?
There's nothing to stop them, no. But that wouldn't change that they have to make the sources available to their customers -and it takes one customer to get the source and provide it for a fork.
But if getting caught making further distribution of RHEL causes a Customer to lose their support contract then the Customers would be unlikely to do so.
I guess what I am trying to say is that perhaps it's an inherently flawed model as evidenced but what just happened and, going forward, we should rid ourselves of such ties.
At the end of the day, we're talking about a super stable server distro with LTS. While many things are easier said than done, I believe this is important enough to the community that it could happen. My main concern with such a distro is security, a team developers providing security patches would be the most critical part of such an entity.
Hopefully before CentOS 7 goes dark in 2024 there will be viable alternatives. Right now, having to compile against custom libs not shipped with the distro (for example, modern OpenSSL) is a small price to pay for having another 3 years left to decide on best course of action to get off CentOS. Still, it's a terrible position to be in for those who upgraded to CentOS 8 already (as we were just about to do).
That answers my question. I had no idea RedHat freely distributed the source to non-subscribers. Of course, they are under no obligation to distribute that to the public.
No, but since they have no cost trial and developer options, its basically easier then managing who gets access to what, as anyone who gets those no cost versions are entitled to the source for those packages.
RHOS is OpenStack. The only two ways two get RHEL 8 sources are from your account’s download page (very tedious) or to use the *-source-rpms repos (reposync/manually) that are available with your subscription.
GPL is copyleft, they're mainly doing additions and modifications to code that is GPL. They have to publish their code. I believe they try to be sneaky and hide how they add it but it's all open source.
The support contract between RedHat and the subscriber isn't anything to do with the GPL. That's a separate document. They could terminate your support if you buy a red car. They can't stop you from buying a red car (distributing the source to RHEL) but they can terminate your support and perform no further distribution of GPL'd software to you.
Terminating support contracts for using your rights could easily run afoul of other laws in different regions as anti-competition. It only takes 1 region to take this approach and the code is out.
Making a support contract dependent on restrictions on distribution sounds a lot like imposing additional conditions on the GPL, just with extra steps.
I don't buy that. They're not restricting their subscribers from distributing the GPL'd code. They're restricting who they choose to offer support to. They're not adding any conditions to the GPL. You can exercise your rights under the GPL. You just may not ever get support from them again.
Actually I think (I don't know that it's been tested in court) that this is a known security vulnerability in the GPL, as long as they only decline to do any further business with that customer (including renewing their subscription), rather than reneg on a already-agreed-to deal.
My relatively conservative organization would never choose a new Linux distribution to replace CentOS, especially not one with a cheeky name. CentOS has a long track record and is well-trusted. I wish the developers all the best but if CentOS is gone, I suspect we’ll be discussing migrating to Oracle Linux (egh) or even Ubuntu. Rocky Linux will not even be discussed as an alternative.
That seems like a surprisingly capricious reason to make a decision that's going to have medium sized real-world impact in terms of rebuilding around a new distros way of doing things.
I am not saying I agree with it, just pointing out that some organizations consider long-term viability as a key component to these kinds of decisions. Just because a fork exists doesn’t qualify it to be a drop-in substitute.
For those who have glossed over the recent CentOS posts, because you don't really know what Stream is, or maybe didn't realize that Red Hat effectively owns CentOS now, this article is an accessible survey.
My question for the Rocky Linux team: if CentOS was struggling for resources before Red Hat's involvement in 2014, how is Rocky any better positioned to succeed?
I am not part of that team, but there are two possible answers: Experience and Motivation. CentOS should provide some experience, best practices, etc., for guidance. CentOS Stream should provide for some motivation, if it really turna out that it cannot be used as a replacement for CentOS 8.
> What will Oracle do for Oracle Enterprise Linux?
Push it as a free installation as widely as possible and then go after the people installing it with some reason they should pay? What should one expect Oracle to do?
OEL is free and widely distributed because it's the environment where Oracle Database runs. But if it gets a life on its own, I do fully expect it will get monetized.
Can you point to some of the free features they've taken back? I have been using VirtualBox nearly every day at work for the past 4 years, and my experience is that if anything, it's improved in quality and free features since the Oracle acquisition.
I do realise how toxic it is to have the proprietary extension pack as the source of advanced features, but the only thing that's I really, really need from that is USB 2.0. It's not like new features are even extension pack exclusives either, most of them land in the open core from the start.
Assuming Centos stream is still having point releases, how different is tracking the previous point release of repos from using centos currently?
e.g. if this is what the tick-tock looks like:
day 1: Centos stream 8.3 is being updated and maintained
day 2: RHEL 8.3 is released based on stream 8.3.
day 3: Centos Stream is now 8.4
After day 3 can you still just track 8.3? Is the issue that centos package repos won’t get updates (particularly security updates) at that point, while RHEL does?
> After day 3 can you still just track 8.3? Is the issue that centos package repos won’t get updates (particularly security updates) at that point, while RHEL does?
CentOS has never released backported updates to previous minor releases. You've always had to stay on the latest minor release, so it's not like there's any real change here.
Right, but the difference today is that Centos is always behind RHEL updates though, yeah? In the future, Centos Stream is always ahead of RHEL point releases, and on day 2 they are close to equivalent, and on day 3/4, centos is goes back to being ahead (at 8.4 and getting security updates), where rhel is at 8.3 and getting security updates.
Assuming that, on the day of a RHEL point release, the two distros are roughly equivalent in packages, with the caveat that the Centos package repos at that point move to the next release (8.4), so you don't get any updates ( presumably mostly security related) in the 8.3 repos... at that point - with those assumptions, if you were on the previous Centos (8.2) - would waiting for the RHEL release to upgrade to the next, now unsupported Centos point release (8.3) be close to roughly equivalent with what the situation is today in terms of compatibility/stability? (I'm not sure what the answer there is)
Or - could just track behind a minor release to approximate the situation today with the caveat that you're not going to get all the updates? Does Rocky linux end up just being the previous centos stream version with updates to the package repos?
I don't use CentOS, but the negativity around this both surprises and does not surprise me.
CentOS seemed like a way to get RHEL for free, lots of people jumped on that and suddenly, when it's no longer free, they throw a tantrum (my favorite: "they betrayed the community"). This strikes me as both ridiculous and just the way FOSS works.
No, it's how a bad way of forcing a monetisation model onto FOSS works. Breaking promises (cutting 80% of the support term of CentOS 8) is not how FOSS became successful.
I cannot comment on the soundness of the decision, but what was the deal with the community here? I can for example see that Debian is community driven with all the advantages and drawbacks that come with this, but was CentOS ever more than a relabeling of RHEL RPMs in the hopes that people would pay for RHEL eventually? I just don't see why anyone can demand anything (not even promises) from RH here.
Why would anyone who has a good understanding of the current software paradigms chose to develop their software for redhat..
I understand that companies like SAP might use redhat. Mostly A political decision made by people who has a say in spendings of to much money.
Cent OS seamed like a bad idea from the start. It does not fill the political reason that allows for redhat. And it does not fill the checkboxes for what a modern independent developer would want.
Maybe I am naive but I can’t see any reason for 1 not caring about the OS and some type of container or “server less” service or just using Ubuntu. Ubuntu is the Granny Smith of apples most things I see is built for Ubuntu and it just works.
1) Nobody ever gets fired for buying IBM. (This quote was relevant here even before IBM literally acquired Redhat).
2) Redhat offers support; and that support is a potentially valuable service worth the cost. If you want that support, you need to use RHEL.
3) If you are developing for a regulated industry, RHEL may be a pre-certified base you can use. This can substantially reduce the cost and risk with certifying your entire system compared with trying to use a not-yet certified base.
As someone who develops for RHEL because of (3), CentOS filled the niche of 'development environment for RHEL without needing to worry about licencing issues'.
It also served as a way out of vendor lock in if you wanted to stop paying for RHEL without needing to migrate your system.
Not every company deals in web apps and streaming videos. Some companies run the NY Stock Exchange, national and international telecommunications, and managing extremely sensitive information.
140 comments
[ 3.0 ms ] story [ 192 ms ] threadIt's funny, when a corporation buy's a community and then decides something without asking the community and even cut the support-time, who could imagine that people get angry?? Has RedHat the same marketing firm as Oracle?
The license has nothing todo with it.
And if you are going to tell me about Red-Hat being built on top of GPL software, well there is Fedora.
Low effort comments can simply be downvoted.
Now CentOS's practical goal - repackaging RHEL gratis - isn't really compatible with what a true community project like Debian is about, in my view. There's no real way for the project to steer away from whatever RHEL chooses to deliver, in neither a positive or negative direction. In that sense, "CentOS Streams", as a new upstream-of-sorts for RHEL, is much more "C"-enabling that what CentOS has been before - because a community (if one will emerge, or maybe it already has) can actually take action to change their own fate, and not totally depend on whatever their Enterprise Linux Overlords decide to make happen.
A pity we're not really looking for this kind of thing where I work now, and most current users of CentOS probably weren't really in it for the "C", but more for the "entOS" part. They were looking for being able to rapidly roll out RHEL clone installations, without having to pay or care about any stupid and impeding commercial licensing restrictions at all. Well, the free lunch at that particular place is over, it seems.
That said, I'm curious to see what will take CentOS's place in that regard.
The community response is in no way FUD. They took a project, killed it, and expect you to like whatever rolling alpha they pretend is it's replacement.
That's as far from "innovation" as you can get - and that was the point. We didn't want the bleeding edge. We wanted the spine of the blade upon which to build our bleeding edge software.
Let's say, Nginx. On Debian, you might be running "buster", their stable release, and running nginx 1.18-1.19. On CentOS, you'd be running on nginx 1.14. This means that you're missing features, but the versions you're running are about as bulletproof as software gets, and (were) still receiving security updates.
Newer versions are available, but usually from the developers directly; the official repos always held much older, stable versions.
When they offered "Decades of Support", they meant it (well, they used to mean it).
Just trust us, we backported that...
Here is the mailing list entry for when they were acqi-hired : https://lists.centos.org/pipermail/centos-announce/2014-Janu...
Instead of seeing CentOS fold, Red Hat decided to hire the core members and support the project. They set up a governance structure for the project, which pretty much guaranteed a RH majority in all votes. AFAICT, there was no requirement for public voting, board seat terms, elections, or public input.
CentOS never really had the strong (public) governance model of something like Debian. RedHat bought the project the old fashioned way — they hired the people who worked on it and stacked the board.
And it's the exact opposite of what I relied upon CentOS for in the past.
In the past, CentOS was what I used when I needed a rock solid OS and software packages that could run for months, if not years, with few manual interventions.
Oh well, I guess there's Oracle Unbreakable Linux to fill this gap until something better comes along.
In the event the CentOS repos get pulled, then I imagine one could do the same thing, but with Oracle's repos instead.
Mind! I don't mean to say that, of my own knowledge, what there is particularly dead about a doornail. I might have been inclined, myself, to regard a coffin-nail as the deadest piece of ironmongery in the trade. But the wisdom of our ancestors is in the simile; and my unhallowed hands shall not disturb it, or the Country's done for. You will therefore permit me to repeat, emphatically, that RHN was as dead as a doornail.”
With apologies to Charles Dickens[0].
[0] https://www.gutenberg.org/files/46/46-h/46-h.htm
I can see the use for such a distribution (run CI against CentOS stream to catch any potential breakage with upcoming RHEL changes) but I think the way they went about making this change is ... unfortunate.
EDIT:
To expand a bit, things may still turn out positive overall if Rocky Linux succeeds and Red Hat doesn't pull off any other silliness. There's an opportunity here for Red Hat to really open up RHEL development to interested parties via CentOS Stream, and so long as they maintain their commitment to publishing RHEL as free software and don't actively interfere with people building their own RHEL-compatible clones from the point releases, everyone wins in the end.
To have an Arch-like distro under the CentOS brand would not make any sense.
I see this misunderstanding all over the net — people think that Debian Testing is more supported than Debian Unstable, when the reverse is true. The Debian project has serious problems with public communications and expectation management.
openSUSE: Tumbleweed -> Leap -> SLES
RedHat: Fedorda -> CentOS Stream -> RHEL
I think most of the complainers don't realize how easy it is to upgrade OpenSuse compared to CentOS. It's like night and day. SLES also has kernel updates with no reboot and an immutable root version. Basically / is read-only and all patches are applied to a new snapshot that is mounted at next boot.
Fedora -----------------------------------------------------> CentOS Stream --> RHEL
CentOS Stream is at most a couple of months ahead of RHEL, not true for Fedora.
So, not a beta for RHEL 8->9->etc, but a beta for 8.1->8.2->8.3.
It's technically rolling release compared to RHEL, but the window of acceptable changes is very small. It's not like Arch and Rawhide and Debian Testing where major OS components can be swapped out under your feet. The majority of the changes would fall into the categories of bugfix backports, backports of extremely useful/important features, and backports of support for new hardware.
Is there a desktop RHEL? If there is, then Fedora may be used to test software, therefore the stability will suffer.
It will only be a matter of time before IBM decide money spent on Fedora is wasted.
CentOS is the new upstream for RHEL minor releases. Fedora is and will continue to be the the RHEL upstream for major releases.
Lifecycle of RHEL:
Those last two will loop until the RHEL full support phase (first 5 years) are over, at which point Stream will end due to there being no more minor releases. Any consumers of Stream will need to migrate to the next major version of Stream.Stream will have effectively 6 years of support (5 + 1 year pre-release), and RHEL will be releasing major versions every 3 years.
Considering new technology and design decisions aren't incubated at all in CentOS, it is in no way shape or form possible to use it to replace Fedora.
* ELN: Fedora Rawhide with RHEL buildroot context.
https://rockylinux.org/
Expect them to change licensing whenever possible and to use more restrictive licensing on new features to make it impossible for anyone to completely fork Redhat Enterprise in the future.
Obviously not related to this case, but seems like a pretty good indicator that the Venn Diagram has at least some overlap.
1. IBM being IBM and forcing RedHat to milk RHEL for all it is worth ecosystem be damned 2. RedHat deciding to do this all on their own without even requiring any pressure from IBM - because they are just as bad as IBM now
IMO #2.
It was not unreasonable to assume that they would maintain CentOS.
This change has occurred now that Red Hat are IBM. I expect to see more, bad, changes. Cutting devs from free software contribution that aren't "directly important" enough, killing free software stacks like JBoss that compete with IBM offerings, and so on.
Enthusiasm, and a bit of counterculture in the mix, seems like a good sign for the long-term health of the organization, IMHO.
BTW, Linux conferences at least used to be very social events, for developers who'd only been interacting online in text, to get together. Maybe people were just really happy to be there.
That giddy thing is also a tradeshow thing, some companies really pump their employees that work at these things, but they're not normally like that :)
Only if you know someone well enough you can get real info. The real communication at these tradeshows is not done in the booths but in back rooms and coffee corners. The front booths are just marketing with some low level PR staff.
I've heard time and again that RedHat will terminate a subscriber's contract if the subscriber is caught distributing RHEL source or binaries to discourage further distribution. Detecting such redistribution seems difficult to me, but setting that aside and assuming it's true, how does a project like CentOS or Rocky Linux get RHEL sources in a sustainable manner?
http://ftp.redhat.com/redhat/linux/enterprise/8Base/en/RHOS/...
At the end of the day, we're talking about a super stable server distro with LTS. While many things are easier said than done, I believe this is important enough to the community that it could happen. My main concern with such a distro is security, a team developers providing security patches would be the most critical part of such an entity.
Hopefully before CentOS 7 goes dark in 2024 there will be viable alternatives. Right now, having to compile against custom libs not shipped with the distro (for example, modern OpenSSL) is a small price to pay for having another 3 years left to decide on best course of action to get off CentOS. Still, it's a terrible position to be in for those who upgraded to CentOS 8 already (as we were just about to do).
I also heard they obfuscated some code and removed comments to make things more difficult, before the RedHat acquisition of CentOS.
Terminating a contract when a customer redistributes the code seems to be a violation of the GPL to me, but I'm not a lawyer.
My question for the Rocky Linux team: if CentOS was struggling for resources before Red Hat's involvement in 2014, how is Rocky any better positioned to succeed?
What will the users of OEL do? I mean, they're all corporations, but still...
Push it as a free installation as widely as possible and then go after the people installing it with some reason they should pay? What should one expect Oracle to do?
OEL is free and widely distributed because it's the environment where Oracle Database runs. But if it gets a life on its own, I do fully expect it will get monetized.
Ok, they didn't go threatening to sue everybody that ever installed it. That's exceeding my expectations already, but it's not a neutral stance.
I do realise how toxic it is to have the proprietary extension pack as the source of advanced features, but the only thing that's I really, really need from that is USB 2.0. It's not like new features are even extension pack exclusives either, most of them land in the open core from the start.
But I do agree it has become more stable and faster.
It’s ahead of CentOS pretty much all of the time.
e.g. if this is what the tick-tock looks like:
day 1: Centos stream 8.3 is being updated and maintained
day 2: RHEL 8.3 is released based on stream 8.3.
day 3: Centos Stream is now 8.4
After day 3 can you still just track 8.3? Is the issue that centos package repos won’t get updates (particularly security updates) at that point, while RHEL does?
CentOS has never released backported updates to previous minor releases. You've always had to stay on the latest minor release, so it's not like there's any real change here.
Assuming that, on the day of a RHEL point release, the two distros are roughly equivalent in packages, with the caveat that the Centos package repos at that point move to the next release (8.4), so you don't get any updates ( presumably mostly security related) in the 8.3 repos... at that point - with those assumptions, if you were on the previous Centos (8.2) - would waiting for the RHEL release to upgrade to the next, now unsupported Centos point release (8.3) be close to roughly equivalent with what the situation is today in terms of compatibility/stability? (I'm not sure what the answer there is)
Or - could just track behind a minor release to approximate the situation today with the caveat that you're not going to get all the updates? Does Rocky linux end up just being the previous centos stream version with updates to the package repos?
CentOS seemed like a way to get RHEL for free, lots of people jumped on that and suddenly, when it's no longer free, they throw a tantrum (my favorite: "they betrayed the community"). This strikes me as both ridiculous and just the way FOSS works.
"Linux is a process, not a product." -- Ian Murdock
I understand that companies like SAP might use redhat. Mostly A political decision made by people who has a say in spendings of to much money.
Cent OS seamed like a bad idea from the start. It does not fill the political reason that allows for redhat. And it does not fill the checkboxes for what a modern independent developer would want.
Maybe I am naive but I can’t see any reason for 1 not caring about the OS and some type of container or “server less” service or just using Ubuntu. Ubuntu is the Granny Smith of apples most things I see is built for Ubuntu and it just works.
2) Redhat offers support; and that support is a potentially valuable service worth the cost. If you want that support, you need to use RHEL.
3) If you are developing for a regulated industry, RHEL may be a pre-certified base you can use. This can substantially reduce the cost and risk with certifying your entire system compared with trying to use a not-yet certified base.
As someone who develops for RHEL because of (3), CentOS filled the niche of 'development environment for RHEL without needing to worry about licencing issues'.
It also served as a way out of vendor lock in if you wanted to stop paying for RHEL without needing to migrate your system.
These companies need to care about the OS.
These accreditations can be had on other systems to some degree but RHEL is the defacto in these realms.
Corporations need us , more then we need them
Do we need to support
- systemd
- gnome
- wayland
- pulse
- copyright drivers in the kernal ?
https://listserv.fnal.gov/scripts/wa.exe?A1=ind2012&L=SCIENT...
https://listserv.fnal.gov/scripts/wa.exe?A1=ind2012&L=SCIENT...
Plenty of chatter in https://listserv.fnal.gov/scripts/wa.exe?A1=ind2012&L=SCIENT...