Slashdot Mirror
Red Hat To Help Develop CentOS
An anonymous reader writes with news that Red Hat and the CentOS project are "joining forces" to develop the next version of CentOS. For years, CentOS has been a popular choice for users who want to use Red Hat Enterprise Linux without having to pay for it. Some of the CentOS developers are moving to Red Hat, but they won't be working on RHEL — they say the "firewall" between the two distros will remain in place. CentOS Project Chair Karanbir Singh said, 'The changes we make are going to be community inclusive, and promoted, proposed, formalised, and actioned in an open community centric manner on the centos-devel mailing list. And I highly encourage everyone to come along and participate.'
Buzzword bingo anyone all in one sentence.
I understand GPL allowing CentOS and Scientific Linux to use Redhat in their respective products, but I find it really puzzling that they would actively *help* CentOS... Doesn't make a lot of sense to me...
THANK YOU, Edward Snowden!! Americans owe you a debt of gratitude (whether they know it or not..)
Sometimes you're really only interested in getting it to work, not keeping it running. Or maybe it's your job to keep it running. If you have other needs, say, science needs, the "free" part is a lucrative proposition when you factor in overhead margin.
...and dumping all of our Redhat licenses. There's no need to pay Redhat thousands of dollars when Centos is the same thing. We already have a mix of Centos and Redhat and the Redhat licenses don't give us anything.
Those of us who've been using Centos understand that if you use it to deploy, and ultimately in your data center, often in place of Windows, then it is just a matter of time before you begin to use RHEL to get support for at least their mission critical production boxes. Centos and RHEL are a nice mix. So, this definitely makes sense for RH. Plus, they have nothing to lose since Centos thrives with or without their endoresment.
Yet, the back and forth relationship RH has taken over the years with the community-driven open source from which it was born and has built its business suggests that, despite this move, they only seem to consider relationships that produce pofits from no more than one degree of separation. This makes the end to this very long estrangement, where Centos only referred to Redhat as the "upstream provider" to keep RedHat's trademark legal team at bay, just plain-old WEIRD.
The question is, how will RH help Centos? That isn't very clear from this announcement.
CentOS 6 was delayed quite a bit from the corresponding RHEL release, for a variety of reasons. If being an unofficial-official Red Hat project means that CentOS 7 tracks the upcoming RHEL 7 release better, then everybody wins. (Conversely, if they turn into Sunacle, then we're likely moving to Debian.)
I use Macs for work, Linux for education, and Windows for cardplaying.
As part of this though, are they going to be moving to an actual open and inclusion development process for CentOS?
I bet they have new customers coming in that started with CentOS. This will give them a little more control over the numbers and I bet it wont be long that a nice red hat rep will contact you after you download and install CentOS. This is a lead generating gold mine for them.
I've been using CentOS and other Red Hat derivatives for 15 years. When I want to get a certification, who do you think I'll get it from? Microsoft? I'll get Red Hat certified, of course. When my employer, a government agency, adds new servers and wants enterprise support, which OS am I going to recommend. Hint - not Ubuntu.
Red Hat isn't competing with CentOS. They are competing with other large companies selling enterprise support, certifications, and training. More people using Linux is good for Red Hat and especially more people being comfortable with Red Hat derived systems is good for Red Hat.
Originally, Red Hat Linux was free. The company was built on cooperating freely with the communityand
contributing, while earning a reputation that allowed them to sell support, training, etc. Working with the CentOS community is classic Red Hat, that's the kind of thinking that once made Red Hat THE Linux distribution and the #3 operating system behind Windows and Mac.
It looks like yes, from the FAQ
Red Hat has worked with the CentOS Project to establish a merit-based open governance model for the CentOS Project, allowing for greater contribution and participation through increased transparency and access.
The Redhat/CentOS kernel is about five years old. Still using version 2.6.
I suspect most desktop users want something newer than that.
To understand this, you have to understand the relationship Red Hat Enterprise Linux has with recompile derivatives. While the compiled RPMs for RHEL cost money and are not redistributable without a license, the source RPMs are nearly all open source. Anyone with a RHEL license can download the RHEL SRPMs and do a recompile. This was great for people who want a RHEL-alike without paying for licenses and CentOS (and then Scientific Linux) came into existence. Red Hat was pleased with this because it gave a cheap way for enterprise customers to try RHEL and eventually become customers who pay for licenses/support.
Then came Oracle Linux who did the exact same thing as CentOS and Scientific Linux, but started charging for licenses and support outside of Red Hat's control. Red Hat wasn't pleased so they started packaging their SRPMs so instead of them containing upstream tarball with RH patch files, they would ship tarballs only or mega huge patch files without comments pointing to the relevent Red Hat bugzilla bug. This made it harder for Oracle to provide support to their customers, but it also had the effect of causing CentOS to get delayed by a good amount every new RHEL release.
Without a quick turnaround on CentOS releases that match RHEL releases, it threatened to kill their "the first one is free" business model. And it probably caused some customers to switch to cheaper Oracle value-added distributors. So Red Hat's only remaining move is to make a relationship with CentOS official. Presumably most of the relationship with be done in private to keep Oracle from gaining an advantage.
I suspect few desktop users run an OS targeted for "servers" where stability is the number one goal?
"As part of this though, are they going to be moving to an actual open and inclusion development process for CentOS?"
No. They get supermajority in the governing board. Red Hat controls the show from now on:
* Ralph Angenent - ???
* Tru Hyunh - ???
* Johnny Hughes Jr - redhat
* Jim Perrin - redhat
* Karanbir Singh - redhat
* Fabian Arrotin - redhat
* Carl Trieloff - redhat
* Karsten Wade - redhat
* Mike McLean - redhat
Quite a clever move. With Fedora they got community approval and support for their betatesting process; with this, they will make possible a flourishing enterprisey open source ecosystem that is menacing going Ubuntu (they hope that, say, the next OpenStack will be "natively" developed on Red Hat). And, of course, they gain traction to be translated into lock in against Oracle and Debian derivatives.
RHEL 7 Beta is based off Fedora 19, with a 3.10 kernel. Usually their beta cycles run about 6 months. Oh, and they heavily backport to their stable kernel (it is "stable", not meaning crashes less, but referring too the fact that the API/ABI doesn't change when they release updates).
2.6.0 came out in 2004, 3.0 (the next after 2.0.39) in 2011. You are not being very precise saying 2.6 related to redhat kernel. But, about to your point, Redhat/centos 5.x came with kernel 2.6.18 (released in 2007, still had the same kernel version in RedHat 5.10 that came out last october), and Redhat 6.x, that came out in 2010, had kernel 2.6.32 (released in 2009). As enterprise distribution, what matters is stability, and certification for 3rd party software, not having the latest versions, all is tested with an specific kernel version, and that kernel (and in general, packages) are kept in the same version, backporting/patching fixes when necessary, and you won't have to worry about a newer version of a sofware changing a configuration file format or keywords and stopping working after updating. Anyway, you can still install extra repositories (like EPEL) that will give you newer versions of some packages.
If you want to use something bleeding edge, you can try Fedora, Ubuntu, or another of the desktop distributions
I have had occasion to run into a kernel bug (back when AMD64 was still a new adventure) and, due to being at a large organization that regularly paid Red Hat various sums of cash, was put in direct contact with high level support. I provided them with my analysis of what was apparently going wrong and a C program to reproduce the failure in a short time (otherwise it would only occur naturally after a system had been running jobs for several weeks to months). Within a day or two they sent back a custom patched kernel that fixed the issue, and later rolled that fix out generally in the next update release (though, admittedly, that second part took quite some time). I might be a competent programmer but diagnosing and fixing a fundamental problem in the kernel and then being on the hook if it has undesirable side-effects isn't something I'd want to do myself, nor could I expect such a rapid answer from the community for what was basically a small corner-case problem, but one that was affecting our business. Having the support is what made the difference.
Of course, for many cases, self support and community support can be good enough, but all it takes is one major issue where you can't solve the problem and you're losing revenue and reputation, and all the sudden that "expensive" support starts to look really cheap in comparison.
And I also agree, stay far far away from Oracle Linux if at all possible (heck stay away from Oracle altogether if you can, but that's frequently impractical).
At some customer sites, I've hear Red Hat has put in clauses forbidding the customer to run CentOS. The clause does not apply to other Linux variants.
Anyone else heard of this? And if it's true, will Red Hat now be softening this stance?
No matter how hard they try, the grandaddy of all distros will always be Slackware.
My understanding is that CentOS-6.4 is 99.93% RHeL-6.4 the .017% being differences in
branding packages and a handful closed-source proprietary 3rd-party packages nobody
ever missed. In my days I have __rarely__ seen a RHEL system and the handfulI I did see had
their support run out and never renewed. Everywhere I've been everybody uses CentOS and we're not talking
startups but big established players. I seriously wonder how RH manages to stay afloat
they must have some pretty big government contracts they're making the Fortune-10 pay out of their
asses, because the money is not rolling in even at the F500 level.
Not only does that just work but so does my approaching 5 year old Windows 7 too!
As a desktop user I do not have to worry about an update killing something because it uses a standard abi like other OSes and unlike Ubuntu throughout 6.x. My scripts will work without breaking, all the apps have matured and are well tested. Driver makers target it, and I keep gnome 2.x and don't have to worry about guis designed for teenagers.
I get a minor update each year! ... Oh and every 5 years I get that huge update 7.x where you Ubuntu guys worked on all the bugs for me :-)
What's to hate about it? I have work to do and do not want to play with operating systems too much.
http://saveie6.com/
I do!
Many others too where we pulled our hair out on bugs and things breaking and being cutting edge. It is updated each year and XP diehards have taught us is that new isn't always better
http://saveie6.com/
And there is a distro for just that purpose:
http://distrowatch.com/table.php?distribution=stella
Personally I probably wouldn't run something like CentOS/RHEL on my primary desktop or laptop since I like to run all the latest stuff without too much of a wait. But if I had a secondary "work" machine and wanted absolute rock-solid stability and unsurpassed support (ten years), then such an OS would be excellent. Running a machine with for the most part only minor updates being required and no major, potential stability-damaging upgrades for its entire working life does sound somewhat appealing if you just want a machine to work, and that especially suits a desktop in the corner that just always works, is always there, never needs any maintenance...
But yeah, I'd probably still upgrade once every other OS version at least anyway. I would eventually get bored and want to start playing with something new.
I don't want CentOS to be eaten and then shat out by Red Hat. I see no long term upside to this. Here is my "Mark these words": Within 5 years CentOS will solely exist as a marketing platform for Red Hat or it will be dead.
Red Hat has piles of cash, CentOS (at best) has piles of pennies. This is not a relationship of equals. I not only use CentOS because it is free but because I don't like the flavor of Red Hat; to me it is the most MS of Linux with Ubuntu a far distant second.
When Oracle snagged MySQL they swore on a stack of bibles that it would be left alone. Yet everyone is now switching to MariaDB, and I suspect that Oracle could make a solid argument that they have been kind to MySQL; the sort of argument that wins high school debating contests but is still a load of crap.
Its actually about time. We old timers remember when RedHat was free and support was the money maker for RedHat. Then they split to RHEL and Fedora, that was bad and caused a lot of initial distrust of RedHat. Fortunately, RedHat didn't screw everyone and is doing largely the right thing.
The problem with the RHEL/Fedora split was it made two different strategies. If it were not for CentOS, RHEL may have lost a lot of business. Now that Oracle wants to steal RedHat business, keeping CentOS viable keeps the mind-share of people who neither need nor want support using the equivalent of RHEL while RedHat keeps its customers.
The big selling point of CentOS was that it was a free CLONE of RHEL. With the talk of "innovation", does this mean that these two distros will deviate and undermine this advantage?
Even Debian that is quite conservative and slow at adapting packages, has been using kernel 3 in production for quite some time. (2 years?). I suspect rather more RH being behind in kernel releases to protect their code investment in heavy customisations to the kernel, which initially was one of the reasons, between many others, that I moved to Debian.
Scientific Linux does a similar integration job, but adds a few critical tools that Red Hat refuses to include as part of their core distribution. These include the "epel-release" yum configuraitons, the "atrpms" repository for acess to the "libdvdcss" library for encoding or decoding DVD's, the "rpmfusion" repo that contains MPEG libraries, and several others. Such repositories are what make RHEL and CentOS *useful* when you need freeware components such as these and you live in a country where they are legal to download. Red Hat, and thus CentOS, won't ever include hooks to these repositories, not even EPEL. If you want to use Red Hat or CentOS for DVD viewing while you're at home, don't bother. Install Scientific Linux and get the "yum-conf-atrpme" package to set up access to the DVD libraries, and "yum-conf-epel" to set up EPEL, etc., etc., etc.
And frankly, the CentOS user group has been getting a bit fussy. Scientific Linux mailing lists have been very helpful, and even developers like Dag Wiers (the creator of RPMforge) has tossed in the towel on CentOS.
The headline is absolutely preposterous, based on utter ignorance. There IS NO CentOS "development" to "help". None. Nada. CentOS simply nabs RHEL source code, debrands it but leaves it otherwise verbatim, and recompiles it into packages with the exact same name and content. Period. It is an excellent way to get an easy to install functional 100% clone of RHEL, and updates, for zero cost, but minus the formal paid support. That is all it ever claimed to be.
One does have to wonder what is ACTUALLY going on here. Presumably Red Hat wants to harness somehow all the energy around CentOS. One suspects the installed CentOS base is vastly larger than the RHEL installed base, and there is a whole lot of energy in unpaid peer support. Presumably Red Hat is eyeing that energy enviously. For CentOS' part, it is much less clear what they gain. Possibly Red Hat gave them an ultimatum, implying or spelling out that they could make their life a living hell, by making it very hard to recompile the source, perhaps as simply as threatening to contaminate the source so thoroughly with Red Hat branding that it would be impractical to "clean" it.
This is all guesswork, but it at least makes some degree of sense as a possibility. Officially, there is absolutely no hint what the motivation is on either side.
Likely the guys at Scientific Linux and the other RHEL clones, the ones that apparently won't be under this new golden umbrella, will have some ideas of substance about what is going on.
The FAQ also admits that Red Hat will now owns a majority of the board members, and the board can only take new members via a majority of the board agreeing.
It's a takeover.
To piss off Larry Ellison!
It's nice to see that Red Hat continues their lengthy streak of hiring assholes by hiring Johnny Hughes.
No, it hasn't even been a year since Debian switched to 3. Kernel 3 (specifically 3.2 which was released 2012-01-04) only came to Debian at the newest release (wheezy) which was released 2013-05-04. The previous version (squeeze) used 2.6.32 (released 2009-12-03) and was released 2011-02-06.
I suspect few desktop users run an OS targeted for "servers" where stability is the number one goal?
Actually, I used to be working on a CentOS workstation for quite some time and it was a very pleasant experience. The only issue I remember is that I had to manually compile some applications to be able to watch soccer streams during EURO 2012, but I'm pretty sure people at the IT department saw this as a feature rather than a bug ...
OS Reviews: Free and Open Source Software
The Redhat/CentOS kernel is about five years old. Still using version 2.6.
I suspect most desktop users want something newer than that.
I suggest that you take a closer look at the acutal kernel source code. The version number is 2.6.32, but that's just because that was when Red Hat _forked_ it from kernel.org. They continiously backport features and hardware support during the production support life cycle.
I suspect few desktop users run an OS targeted for "servers" where stability is the number one goal?
It's not targeted at servers.
RHEL Server is targeted at servers.
RHEL Desktop and RHEL Workstation is targeted at desktops/laptops.
As enterprise distribution, what matters is stability, and certification for 3rd party software, not having the latest versions, all is tested with an specific kernel version, and that kernel (and in general, packages) are kept in the same version, backporting/patching fixes when necessary, and you won't have to worry about a newer version of a sofware changing a configuration file format or keywords and stopping working after updating.
That's true for most software in RHEL, but not the kernel. It is heavily updated in each point release. Take a look at the release notes some time just to see what gets in there. RHEL 5 uses the 2.6.18 kernel for example, yet supports both KVM and ext4; features that Linux didn't have at that point.
The RHEL/CentOS kernel does get security and bug fix backports from later kernels, but the reason it runs such an "old" kernel is for stability reasons. Most Windows desktop users never upgrade their OS to a newer major release during the lifetime of their PC (because it costs money and can be a hassle - remember most of the world's desktops are running the OS that was pre-installed when the machine was bought), but apparently most Linux desktop users are constantly chasing the bleeding edge if you're to be believed.
The problem with most Linux distros is that their support window is very narrow - usually less than 18 months and definitely less than the lifetime of a typical desktop PC. This is where CentOS scores heavily against Ubuntu - 10 years of updates to the OS, so *you* decide when to jump to the next major OS release and you're not effectively forced to jump releases half-way through your PC's lifespan.
Also note that it's not the kernel version that's a problem with CentOS, it's the older system libraries (particularly glibc, X11, Gnome etc.) that cause issues, particularly if you want to run closed-source binaries. Google Chrome is probably the highest profile casualty of this, which is why I cooked up a script to install the latest Google Chrome on CentOS 6.
The RHEL/CentOS kernel does get security and bug fix backports from later kernels, but the reason it runs such an "old" kernel is for stability reasons.
It isn't stable. The kernel is heavily updated in every point release including actual features, not just bug fixes.
There's no reason not to. Desktop users want "stability" too! CentOS/RHEL users can simply add some more repos to get more recent software packages.
The only reason you might avoid a "stable" release, is if you have newer hardware that isn't supported by the old, "stable" kernel and supporting software. Personally, I was able to manage that just fine by compiling my own Fedora 3.x kernel SRPM on CentOS6... It might sound like work, but it's a lot less effort than living with all the horrible nasty bugs that come with a Fedora distro.
Slashdot gets worse every day... Pipedot: News for nerds, without the corporate slant
The only reason you might avoid a "stable" release, is if you have newer hardware that isn't supported by the old, "stable" kernel and supporting software. Personally, I was able to manage that just fine by compiling my own Fedora 3.x kernel SRPM on CentOS6... It might sound like work, but it's a lot less effort than living with all the horrible nasty bugs that come with a Fedora distro.
It has been said multiple times in this thread already. Red Hat continuously updates the kernel with every minor release. There are some exceptions but it will run on modern hardware.
The only reason RH is behind is they haven't had a major release since 3.x came out.
I'm starting to think GNU is the problem with "GNU/Linux" these days.
It might sound like work, but it's a lot less effort than living with all the horrible nasty bugs that come with a Fedora distro.
What nasty bugs are those? I'm a Fedora user and I haven't ex ^H^H
I feel your pain. I wish Fedora would go to a 9 month update schedule, it would make me happier. When I jumped to Linux on x86 from PPC (long story), I narrowed down my distro choices to either CentOS or Fedora. I chose Fedora...maybe I should have gone CentOS but I haven't had too many major problems with Fedora.
Recompiling a kernel can be "guru" work...yes I've done it by following a "recipe" in a book but I haven't done it in years, forgot how.
I refuse to think that RH is into it, but to my paranoid mind, this sounds like a weird variant of the well known MS strategy. I really hope I'm wrong.
Open Source Network Inventory for the masses! Kuwaiba
Fedora moving to a 9 month release schedule would slow down development, which is the exact antithesis of it's intended purpose. Fedora is SUPPOSED to be bleeding edge, not everyday use. It is a crazy testbed by design.
I posted this on G+, but figured I would post it here too.
---------------
As a user of RHEL and CentOS, I use each for specific purposes. Several of the purposes I use CentOS for, Redhat wants me to use RHEL for. Now they could easily make CentOS less of an acceptable option almost forcing me to use RHEL since I use RH as my standard infrastructure distribution.
CentOS has always taken revenue away from RH where a full blown license of RH wasn't necessary and RH knows this. I have no doubts this move is to strengthen RH position and profitability. From what I read in the FAQ about this joining, Redhat requires that Redhat Employees have a controlling share of the CentOS board.
In the end, we will just have to see exactly how Redhat intends to steer CentOS.
We just substituted CentOS for Ubuntu across all of our desktops in the last year. Most users never pay much attention to which kernel their using, unless they are doing device development. What they do notice is that Unity sucks and that CentOS lets them use their beloved Gnome 2 environments until 2020. ( They'll likely migrate off before then, but at least they won't be forced. )
Anyway, you can still install extra repositories (like EPEL) that will give you newer versions of some packages.
I don't think EPEL provides "newer versions of some packages." As I understand it, EPEL only includes packages that do not exist in RHEL at all.
http://arstechnica.com/information-technology/2014/01/red-hat-and-centos-become-voltron-build-free-operating-system-together/ sayus it will be a Voltron together! :P
Ant(Dude) @ Quality Foraged Links (AQFL.net) & The Ant Farm (antfarm.ma.cx / antfarm.home.dhs.org).
MySQL was absorbed by Oracle with similar hopes and aspirations... and look what happened. As I recall, RH was getting really pissed off at CentOS, enough to change their development processes to make it more difficult to re-assemble the distribution. Now, they want to be best pals? Okay, we'll see :-)
Genuinely, the combined effort would be useful to everyone, that is if RH's intentions are as on the level as they claim.
Myself, I'd absolutely love to run RHEL or CentOS on a desktop if I were using Linux on a desktop. You don't have to worry about things becoming fucked up with upgrades. For the few things I test on Linux, I always run CentOS in a VM, but my daily desktop runs on OS X.
A successful API design takes a mixture of software design and pedagogy.
Parts of 3.x is already in the 2.6.32 RH ships.
As long as we are on the topic of Cent. Anyone know the popularity / status... of Scientific Linux. Scientific Linux has some differences in RPMs so it is a bit further away but I'd think that they would want to support it if they offering a support / conversion service. Anyone have any insight?
By helping CentOS, Red Hat will ensure RHEL's future even more. This will also help CentOS as well, which is an excellent choice no only on the server but on the desktop as well. In the end, both parties should benefit. This is definitely good news and also keeps open source GNU/Linux succeeding in the real world.
This is exactly the reason that I have a workstation in our lab running Fedora instead of CentOS. I'd prefer to use a distro closer to RHEL, since that's what run some of our expensive instruments, but the drivers just weren't there and it's really not worth my time to sort out all the dependencies or compile a newer kernel without breaking other stuff in order to get support for newer hardware.
I'd also like to know what nasty bugs you're talking about in Fedora releases. I've had a few non-critical packages break for a week or so after an update, nearly all of them in rpmfusion and not the core Fedora repos, but that's about all.
It has a lot of backports.
Opensuse 13.1 is no less stable than RH/Centos.
Scientific Linux is another RHEL clone like CentOS. It's more desktop oriented.
If you run Google Chrome, RHEL 6.x based systems are not supported because the libraries are too old by Google's standards.
I feel your pain. I wish Fedora would go to a 9 month update schedule, it would make me happier.
We probably won't go to 9 months permanently, but it's very likely that Fedora 20 -> F21 will be along those lines as we retool for Fedora.next ideas, and work on improving qa and releng automation.
Not true with an SRPM package... A vanilla kernel can be a lot of work, but recompiling an SRPM with rpmbuild isn't difficult at all.
Slashdot gets worse every day... Pipedot: News for nerds, without the corporate slant
> Meh.. Windows Server, RHEL... who wants an OS. The VM infrastructure below, the storage, the network,
Yeah nobody uses Linux for virtualization, storage, or networking. Who's ever heard of OpenStack? Well, noone except the major datacenters. And the minor ones. And people at home.
good Centos Linux thanks Red hat
I guess I am a desktop user of CentOS -- but I don't really have a choice, because all the machines on the shared filesystem where I work have to be CentOS or else IT won't support them.
Having stability in our environment is clearly a priority and we have to support a ton of legacy software among the various labs that use the filesystem. Even so, by now the system is too old and probably causes more headaches than if we were using debian stable. "What do you mean, this package requires a newer version of boost than GCC 4.4 supports? Nope, sorry, build it yourself without root."
Last time I looked on a report regarding enterprise market ( some stuff from IDC at work, I cannot publish it, so feel free to doubt my words ), I was surprised to see that Canonical was mentioned only as the smallest player in the server market ( but this was in 2011 market share ). They were growing, but growing fast is easy when you are small ( getting 2 new customers is a 50% increase when you have 4 of them ). And seeing the data, it seems that Canonical mostly managed to steal marketshare from Suse and Oracle, ie, people who already were going to take the cheapest option rather than RH.
And for the vast majority of RH people, I can only hope this would be the start. The problem is that, if you are quite passionate in your role as a project leader, then you have to dedicate full time to it, and either your company allows it, or you would be hired by a sponsor of the project. IE, people who cannot work full time have trouble to match the level of dedication of those that can, and those that manage to be as dedicated would likely just be hired so they can do what they like as full time people. So the choice is between having a diverse set of people, or having set of people that can work full time. There is surely way to have both ways, but that usually not easy, and if people wanted to pay for a centos member to work full time, it would have happened in the past already. ( or it could happen now without trouble )