CentOS Linux is ending because Red Hat refused to invest in it
theregister.com
theregister.com
So the people for whom Red Hat provided no measurable value beyond the intrinsic value of what you downloaded. The copyright for which is largely owned by Not Red Hat, and thus they're bound by the GPL.
Red Hat adds incremental value to those projects and organizes the distribution - they deserve to get paid for that, of course. But they were paid - they are the stick by which all other open source companies measure their commercial success. If you advertise being open source as a good thing, the ability to fork is a feature, not a bug. So the Rocky Linuxes of the world are inevitable and a good thing.
Acquiring the CentOS project and then killing it is destroying a lot of community good will and I have no sympathy...
First of all they were one of the first Linux distro (I remember installing 4.2 at home) with Debian and Slackware. At those times making Linux distro was a highly unprofitable business. Companies did not care about Linux at all. Windows, Sparx, AIX, IBM mainframes, etc. everywhere.
Yet Red Hat people had a vision that Linux could be a thing in the real World and I would say, they deserve a premium for all those hungry years.
Second, Red Hat is one of the largest corporate contributors to Linux kernel. I bet that, for instance, Amazon AWS is making way more money on Linux and open software than Red Hat, yet their contribution is not particularly significant.
I am not surprised that Red Hat is not supporting financially something that is a rip-off of their software. A legal rip-off obviously and Red Hat is not suing or spreading FUD about such initiatives, but I don't see any reason why they should financially back them.
Actually, for a period,in the dot-boom years, Linux distributions took plenty of investor money. Caldera, anyone? RedHat was "just" one of the few to figure out ways to survive the end of that wave.
I think the IBM Redhat purchase could have been a very good thing if handled properly. CentOS was actually a feature of RHEL in a way, but no more.
Unpopular opionion but i think Redhat probably just extended the lifespan of the centos distribution instead of killing it straight away.
(I am saying that as someone who has donated to the rockylinux project)
Based on what? I just went digging through the archives (granted I only spent ~10 minutes) because that's not at all how I remember it, and nothing indicates the CentOS team were facing insurmountable financial hardship or asking the community for more support. Either as part of the acquisition announcement or in the months prior to it.
https://lwn.net/Articles/579551/
https://redmonk.com/dberkholz/2014/01/10/red-hats-centos-acq...
I’m sorry for the CentOS users, but I think the reality is that it’s not a project that many want to work on. It seems to require a lot of resource and it’s an operating used by people who specifically don’t want to pay.
I can get the workstation offering for $299. And then I can support free-as-in-libre software, I can run the same system my customers are running in production, and I get support that's probably better than Microsoft's. I don't really know what device issues I'm going to run into, though, and in my experience it either works or it doesn't - if I have a new wireless device they don't support, I'm under no illusion that my $300 next year will be the difference between them supporting it or not.
So maybe I'll get the self-supported option. But that's kinda steep - over half the cost of full support. Especially when one way to be confident about device issues is to go install the community distro (Fedora) for free and see if an old release works. If it doesn't, I'm not betting on Red Hat. If it does work... do I want to start this whole process over again, or do I just want to get to work.
Well here's another deal breaker: if I'm trying to run the same distro my customer's use in production, I'll have lots of virtual guests. No can do: my workstation license only permits 1. Let me get the server license and just install KDE myself. The self-support option only allows physical hosts. Let me get the next tier up. Now I'm up to $800 a year, and I, as a developer of software for Linux, am going to have a very unusual use case for them, I think. Should I commit $800 for 1 year of this? Does the server edition run KDE well? I doubt many people are trying that. Now I'm too nervous to cough up $800 knowing it only lasts a year anyway.
I could see myself buying a Lenovo laptop with Red Hat pre-installed and skip this, but I would want to be able to run more than 1 virtual guest with the same system - I'm not sure if Lenovo customers get special terms. Or... I can just buy a laptop I think is better from System76 and use Pop! OS. As long as there's something resembling CentOS or Scientific Linux or Rocky Linux for virtual guests, I can get close enough to my customer's production distro that this is still by far the most attractive option.
But personally for me the moment's past and that offering is still not actually available AFAICT. I no longer work with customers in their ecosystem much and the CentOS dealings feel too close to embrace-extend-extinguish for me to feel like championing them again.
It is a difficult situation in general and it is a shame that this was one of the most used distros in servers.
Funding it to get a pathway to onboard RHEL-like workloads without the costly licence? Worth the change they put down for it.
There's a lot of reason to distrust Amazon, but this isn't what ended up happening with Elasticsearch and Kibana. Ultimately, the company Elastic took open source projects, moved them to more restrictive licenses, and Amazon came in and committed to maintaining proper open source forks.
https://blog.cloudlinux.com/announcing-open-sourced-communit...
Also this is a sentimental name, chosen well imho.
> "Thinking back to early CentOS days... My cofounder was Rocky McGaugh. He is no longer with us, so as a H/T to him, who never got to see the success that CentOS came to be, I introduce to you...Rocky Linux"
> difficult or uncertain
> unable to balance very well
https://dictionary.cambridge.org/dictionary/english/rocky
The parent to your comment is correct, as are you about the specific reference in this case.
The etymology of the Rocky Mountains is unrelated to either:
> The Rocky Mountains were so called by 1802, translating French Montagnes Rocheuses, first applied to the Canadian Rockies. "The name is not directly self-descriptive but is an approximate translation of the name of the former Native American people here known as the Assiniboin .... The mountains are in fact not noticeably rocky"
In attempting to research who Rocky was, I stumbled onto this post with a dead link[0]. Anyone know the story behind Rocky and TeamHPC?
[0] https://www.linuxquestions.org/questions/general-10/news-155...
The homage seems lovely, but doesn't (unfortunately) change the fact that to the vast majority of english speakers "rocky" means unstable/dangerous... In fact, one of the most common expressions referring to an unstable/bad relationship is "their relationship is on the rocks" or "they are on the rocks".
People aren't going to think Rocky Linux is unstable for the same reasons why they don't think Apple computers rot if you leave them in the sun. Also, if a product called Red Hat Enterprise Linux was successful despite the name, I'm guessing it can't be that much of an obstacle for Rocky, either.
My reading of the name "Rocky Linux" is similar to my reading of the name "Rocky Balboa". It could be just a name, like "Foo" or it could signify toughness, as "Rocky" does as a nickname in many cases. In two non-fiction cases of boxers, "Rocky" is a diminutive of "Rock" or "The Rock".
Businesses aren't your friends, repeat it with me, no matter how much propoganda they regularly spew out. Sometimes their goals may be temporarily aligned with yours in which case you can leverage their resources but never forget their goals can shift to be misaligned with yours at any point, making them a partial or direct enemy of your goals.
A business exists to make money and wealth for someone or some people. If that's not you, then they're not there to help you. I should point out that there are classes of businesses that do exist somewhat as an exception, certain charities and non-profits, but those aren't representative of the vast majority. Even they often aren't necessarily there to help you per se.
Be wary of either group selling them as such.
If there is anything that poses a risk to the company, interferes with an employees ability to serve the company, or in any way reduces productivity... there will be no difference between canonical and oracle or even canonical and McDonalds.
The difference between your experience working at these companies is largely window dressing on the same corporate employer/employee model.
Again, this is like saying that if you push a person hard enough they will commit violence which immediately implies that all people are violent criminals. It’s simply not how that works, and equating all current and past behavior to what someone is capable of is silly.
Customers, employees, partners etc are all means to one end, and one only - making someone rich
Robbery and murder have no upsides to a society. Organized labor to create something bigger than a single person can build is foundation of society. You can argue which system of organization is better (government owns and dictates everything like communism, corporations and government own and some specific categories of things and dictate what’s in their purview like capitalism, the royalty and nobility own and dictate everything like monarchy, or no organized labor exists like anarchy). But the reality is that we as a species tend to organize into tribes and work in groups and whether you call those tribes/groups corporations, government work forces, peasantry, guilds is just about what trade offs you want to make.
I will be really impressed if you can show a proof that a corporation is like murder.
Also why do we even give credence to ideas such as good corporate “citizenship”? Thank you for not destroying the economy or societal fabric..? Way to be a good for profit entity that probably still isn’t raising wages or helping with the homeless crisis..? There should be strong laws and regulations that bind companies and jail their execs if they step out of line. We shouldn’t depend so heavily on so called good corporate “citizenship”.
I didn’t vote for these executives, their boards, or their investors. This isn’t what democracy looks like.
Look at someplace like Costco- highly profitable, pays well above average for the sector, employee and community friendly policies... which would all go in the trash if Sams Club starting eating their lunch and their margins disappeared.
As fro democracy there's no reason you should have any say in how these people exercise their freedoms under the law, beyond your say in government and the law. If you have a problem with anything they do, then that's a matter for the law and regulation, applied equally to everyone. An idealistic view I accept, but it's what we should be aiming for.
Right up to the moment where there isn't.
Some companies have great values, a great culture, an inspiring mission and they'll be relatively consistent in maintaining these things for quite extended periods of time: years, sometimes even decades. That's really easy to do during the good times: when you're hitting your numbers and all the trendlines are pointing upwards.
However, when things get more difficult - when there's a downturn, competitors move in, or the market simply moves on - that equation changes a lot. Sooner or later under those circumstances every business has to start making hard calls that will make it very clear to all involved that it is not a family.
We can all point to examples of businesses that have gone out of their way to look after their employees during the pandemic. But even they can't do so indefinitely (hopefully they won't have to try if the vaccination programmes work as well as we all hope). Sooner or later the values of a business will become subservient to commercial reality even if that doesn't happen to our values as individuals.
I am not saying businesses should have a carte blanche to do what they want. Far from it. But in the sense that some corporations tend to behave better than others, just like some individual people tend to behave better than others, we should recognize that it is better to push all to behave better than to get rid of all of them on the grounds that the worst ones are beyond redemption.
No, but its a helpful fiction as far as it makes everyone more convivial and agreeable. It forms a kind of shared illusion, not unlike the kayfabe of professional wrestling.
For the uninitiated, B-Corps "must receive a minimum score from an assessment of 'social and environmental performance', [and] integrate B Corp commitments to stakeholders into company governing documents".
Patagonia, for example[1][2], seems to be a trailblazer in being transparent about and actively improving their business in terms of logistics (monitoring "the cost of materials that have the potential to be Fair Trade Certified against the volume of materials that are purchased in a Fiscal Year"), corporate culture ("[encourage] work-life balance, as well as a focus on an individual’s physical and mental health"), end products (their "criteria for the best product rests on function, repairability, and, foremost, durability"), and so on.
[0] https://en.wikipedia.org/wiki/B_Corporation_(certification)
[1] https://bcorporation.net/directory/patagonia-inc
[2] https://www.patagonia.com/stories/benefit-corporation-update...
Unless you work for the family business ;)
Not even sure about those. Quite a lot of charities and non-profits set up to fix a problem end up perpetuating it because it is a nice comfortable job.
I mean one of the richest families in NL have their money in a bank that is part of a charity.
https://www.fastcompany.com/3035734/ikea-is-a-nonprofit-and-...
Essentially, Ikea is profitable because it utilizes a tax-avoidance strategy involving the use of a non-profit and so your comment is in my opinion basically true and they are "cheating" a bit. A pretty slick cheat, though.
Also, thank you for shining a light on this as I was ignorant of the whole thing.
- corporate-driven open source was never about user/developer freedom
- companies have marketing departments and know how to do astroturfing
- any project that follows corporate decisions is not community-driven, even if the contributors are volunteers
Undermining would be if RedHat did things to actively sabotage CentOS, spread untrue FUD about it, file lawsuits, etc.
You can weaken through direct attack, withdraw of assistance, or you can even strategically offer assistance to weaken (identify largely unknown harmful processes and help them grow), amongst an array of other strategies, direct and indirect.
The grey area is if you've been a bystander to action and refuse to act one way or another because intent is the only information you have and you can't (fairly) gauge on that alone. This isn't one of those cases, this is a clear case of withdraw of active assistence which undermined CentOS. Business processes often work in reverse of what is explained. It's not likely Red Hat made a decision and then informed CentOS there would be consequences. It's far far more likely Red Hat looked for decisions that had the consequence they desired, chose one, and then informed CentOS that wasn't their intent.
These sorts of games of misdirection of intent and responsibility are played by businesses and sociopaths all the time. Decisions like this are highly strategic and implications are discussed throughly, meanwhile the narrative is crafted to then look benign and non-strategic.
> What was the specific change? It seems that Red Hat said it would invest in Stream but not CentOS.
CentoOS Stream is still CentOS to me. Better to say Red Hat will not invest in CentOS point releases, as that is a lot less confusing.
What I see here is Redhat optimizing the release cycle. Please correct me if I'm wrong or my english is bad but it seems to make a lot more sense from a development standpoint to have CentOS Stream than the overhead that was each CentOS release.
And in that process they gain a sort of dev branch of RedHat that people can use to get a distro as close as possible to what RedHat will be, but also beta testers.
For many scenarios it does. It gives far more visibility into what's coming in future RHEL releases than a downstream rebuild of past point releases did. However, a lot of people are used to those backward looking point releases because "that's the way we've always done things."(tm)
I also imagine that many of the same people think nothing of using cloud platforms and services that are getting changed many times a day--but those changes are invisible for the most part so it doesn't bother them.
I'm not sure beta test is all that meaningful a term in a CI/CD world. Yes, there is final acceptance testing for official releases but a lot of software we use is basically changing all the time.
But the point releases will be still there - but only for RHEL.
The sad thing, which makes Stream unusable is, that RHEL does rebase packages between point releases. With the old CentOS you knew, that if there are breaking changes, they would be during point release upgrade. With Stream, you will get them randomly (and they did already happen several times since announcement).
What manner of breaking changes would these be?
AFAIK (may be wrong, please correct if I am) by default RHEL is not locked to minor release versions (i.e. no subscription-manager release --set 8.x), but to major release versions (i.e. 8), and I have always used RHEL like this and never had serious issues with minor version updates.
In this sense CentOS stream is really just what I was already doing, CentOS stream is also split by major versions AFAIU (there is 8 stream, and will be 9 stream)[1].
EDIT:
[1]: https://centos.org/distro-faq/#q6-will-there-be-separatepara...
Poppler is going to be rebased from from 0.66.0 to 20.11.0. If you have anything linked to it, it will break; you need new binaries linked against the new soname. There is a surprising number packages linked against it, and third-party ones too (e.g. GDAL, used by all geo apps out there).
Similar thing will happen with other rebased packages.
RHEL users will get that packaged in 8.4, with release notes. Stream users are getting these breakages randomly.
It would be interesting if there was a way to determine if any other open source projects have this kind of large dependency on a single company. Track GitHub accounts doing the commits which are linked to orgs perhaps?
The short answer is that the vast majority of open source projects are to a large degree dependent on a single person or company. (And those without corporate funding can have trouble--as did CentOS at one or more points in its history before Red Hat picked up the funding.)
ADDED: Although major successful projects do tend to have a more diverse community because that's what takes most advantage of the open source development model.
On the other hand, Fedora is upstream of RHEL. So, it serves an important role as a testing ground for stuff that may or may not end up in RHEL later down the line. I'm pretty sure you're safe, as it actually fills an essential role that is useful to Red Hat and that is very unlikely to eat into their main business.
But some are much more likely than others, and choosing which to base yourself on is a form a risk decision; perhaps a minor one.
I wonder if they’ll now switch back to RHEL.
CentOS was TOO GOOD after so long; even C-level execs were happy to sign off on using it.
https://www.zdnet.com/article/red-hat-the-first-billion-doll...
« we went to the CentOS project and said, here is a thing Red Hat is going to do. We believe there are consequences of this action... the end result was the decision that got made by the project »
Something doesn't seem right, if CentOS is really a project capable of making its own decisions as he's trying to make it appear.
If that was the case, I wouldn't expect that the first thing the public to hear about the change would be the announcement that CentOS-in-its-previous-form was being discontinued.
I would expect to hear the board saying « Red Hat say they're no longer going to invest in CentOS in its current form. If nobody comes forward to invest with money or staff, we'll have to discontinue it. »
If you want to use a free distro, why not just move to one of the many distributions which are not tied to a for profit company?
Which means you can set it up once, install your mail server or whatever, and have a solid and secure system for a very long time.
Here’s the crux of it.
"The next version will not be like Stream 8. "We have said publicly this is not the way it's going to be. Stream 9 is going to show you that code flows from Fedora to CentOS Stream"
To me, this indicates Stream will be rather close to Fedora. Right now Stream seems rather stable. But that tells me Stream will be a rolling release of Fedora. Do others read it that way ?
The cognitive dissonance in assuming every acquisition of a smaller company by a bigger one immediately results in the collapse of said small company is unreal.
To quote that page "Once customers upgrade to IBM?s product it is often too late to negotiate and avoid hefty licensing charges associated with changed licensing metrics? "
There's a very good possibility that Linux wouldn't have had the impact that it has had absent IBM's investments in the early 2000s. It's not an "open source company" but it has significant investments in open source.
then again, I see open source slowly but steadily losing ground. and if open source has a tough time. free software has it even worse.
Turns out not compromising your principals and being zealots works.
A lot of OSS development is now done by companies who don't care that much about the software, because their revenue is totally independent. Netflix, Uber, Microsoft, etc.
Or it's open-core like Redis and Gitlab, with paid proprietary bits. Or "open source" with anti-service-provider licenses like MongoDB, Elastic, Redis.
I thought copyleft licensing has been steadily losing share to permissive licensing for a long time, with no end in sight. Can you point to information indicating otherwise?
The Linux kernel itself keeps going from success to success, but that seems to be the last major copyleft bastion, and they also somewhat famously rejected GPLv3.
From Linus[1] on the topic of GPLv3, he chose GPLv2 as it was the best choice for doing kernel development - giving code and getting code back in return. One of his gripes about GPLv3, IIRC, was the 'Tivoization'[2] additions. He felt it was going too far the other way.
[1] https://www.youtube.com/watch?v=PaKIZ7gJlRU [2] https://en.wikipedia.org/wiki/Tivoization
Here's something I wrote at the time: https://www.cnet.com/news/linux-to-gplv3-a-practical-matter-...
CentOS brought people in to the EcoSystem as an alternative to Debian Based distros and that is sadly no more.
EDIT: added "the new" to qualify the statement about RedHat, as indeed it's IBM at play. No amount of "kernel support" is worth the trouble of dealing with them.
No. Suse still exists, if you don't like Debian based distribution. But my money is on the Debians - and more specifically, on Ubuntu.
For server stuff, Ubuntu is great - commercial support with various cloud providers, lots of prepackaged software (including the commercial software I use).
For hiring, a lot of students run either Debian or Ubuntu on their own computers - meaning I don't have to teach them the intricacies of rpm like apt search is rpm -Q, apt install is rpm -i ...
Icing on the cake is ZFS as the rootfs filesystem out of the box since the last few Ubuntus.
RedHat was innovative, like when they replaced ext4 by XFS as the default filesystem, but that was like 15 years ago.
Still, big money make miracles happen. If IBM can arrange some relicensing of ZFS with Oracle, they could use it on RedHat, but it'd take me a lot of cajoling to even consider switching back - much more than a filesystem that's the minimal prerequisite to just put RedHat on the list of options. Bitten once, twice shy you know?
Anyway, given how it's favored for WSL, Microsoft will eventually fork or buy Ubuntu from Canonical or just buy Canonical outright, which will make it even more of a rival to RedHat in the long term.
"rpm" is not an equivalent to "apt", it's an equivalent to "dpkg".
The tool you would actually use is "dnf" or "yum", which isn't any more difficult to use than "apt" is.
"dnf search"
"dnf install"
"dnf remove"
"dnf list"
etc.
It's fine to prefer Debian / Ubuntu but please don't repeat information that's a decade out of date :)
Fair enough!
Canonical has a wide audience but they've been unable to actually translate that into a self-sustaining business. They're lucky that Mark Shuttleworth is committed guy and not too attached to his stacks of money.
My original comment was exactly about the ways canonical, SUSE, Debian and others are dependent on RedHat. Canonical attempted to develop upstart as their in house systemd alternative but they failed. Exactly the same thing happened with their attempt at their own graphics stack.
I don't know a lot about SUSE, but iirc their distro is also systemd based.
Edit: Suse's package manager is also based on RPM.
They also do plenty of kernel development -- not as much as RH due to sheer size, but still significant. Whereas Canonical doesn't really do any.
Plenty if differences.