dcsimg

Send in the Clones: The Long Wait for CentOS 6

Forget Godot, we're waiting for CentOS 6. If you hoped to have a shiny new CentOS release under the Christmas tree, you were disappointed. 2010 slipped by, still no release. Hoping to surprise your honey with CentOS 6 on Valentine's Day? Maybe. If you're lucky.

Red Hat Enterprise Linux (RHEL) 6 was released in November of last year, but where’s CentOS 6? Clone cousin Scientific Linux 6 is in beta now, but CentOS 6 is yet to be seen.

If it seems like an unusually long time to wait for a CentOS release to follow RHEL, it is. The last major release (RHEL 5.0) came out March 14, 2007. It was quickly followed by a release of CentOS on April 12. Scientific Linux 5.0 came out on May 7th — a bit slower, but Scientific isn’t just a rebuild of RHEL, it also includes extra applications and customizations.

Were things different with the 4.0 series? Scientific 4.0 was released on April 21, 2005. CentOS 4.0 was released on March 2, with a AMD 64 release following by a week. Again, not quite a month after RHEL 4 — which came out on February 15. (How considerate of Red Hat to wait until after Valentine’s Day…)

So there does seem to be some foot-dragging going on here. What gives? Something else that’s curious, a distinct lack of public alpha or beta releases coming from the CentOS camp. Scientific Linux is now on its first beta (pardon the hideous Web interface for their mail archive), but CentOS? Nada. During the 5.0 cycle, CentOS had at least one beta out before the public release of RHEL 5. This time around? No such luck.

The CentOS team took a different strategy with this round of releases — instead of releasing betas and trying to shadow Red Hat closely, they directed their community to the Red Hat beta process. That’s not a bad decision, necessarily. Since CentOS is repackaging RHEL there’s no good reason to duplicate testing. Bugs in upstream RHEL software ought to be reported to Red Hat, let Red Hat benefit from the additional testing community and do what its enterprise customers pay it to do (among other things): fix bugs and ship solid software.

When the topic of the long wait has come up on the centos-devel list, the developers have been a bit tetchy. Understandable, in part, because some of the queries have been less than diplomatic. I say “in part,” however, because CentOS could be a bit more inclusive in terms of soliciting outside help and testers. Reading the lists, it’s my impression it’s not necessarily easy to jump into CentOS contribution.

But many in the CentOS community also lack an understanding of what the developers do and why producing CentOS is any more difficult than simply rebuilding the Source RPMs (SRPMs) that Red Hat releases. So let’s take a look at what CentOS contributor R P Herrold lays out as necessary to produce CentOS. In short, it’s more work than meets the eye. Skimming the lists, it also looks like RHEL 6 is new territory in terms of setting up the build infrastructure and installer, which is also taking time.

5.6 or 6.0?

Another problem for CentOS is that the number of RHEL releases are growing. You see, on top of the shiny new 6.0 release, there’s a new RHEL 5.x release (5.6) that’s of interest to the folks who already have CentOS 5.x installed.

Given limited resources in terms of build machines, developers’ time, etc., which one do you push to finish? The CentOS folks are going with 5.6, with 6.0 to follow eventually after the ISOs are ready and the mirrors have had “a bit of a break.” That was the word as of January 14, and there doesn’t seem to be a final release plan in place as of yet.

You Pays Your Money…

For users like myself — those who want to keep their technical chops up to date, or use CentOS to test scenarios on a RHEL 6 clone, the wait is an annoyance but not fatal. Ditto admins who work for businesses that deploy CentOS rather than RHEL. Few businesses are going to be rolling out production deployments based on RHEL 6 already. If anything, RHEL 6 final simply started the process of planning to deploy the 6.x series for most businesses and you won’t see a lot of production use of RHEL 6.x until 6.1.

And ultimately, this one reason why organizations choose to pay the Red Hat subscription fees or choose a distribution like Debian or Ubuntu. If you need to depend on releases and security updates from an organization, you probably want to get it directly from the source. With an intermediary like the CentOS project, there’s going to be some delay involved. Maybe a little, maybe a lot. There’s no accountability from the CentOS project — it gets released when it gets released. Don’t like it? Tell it to someone who cares, or spend the time diving into the release process and help fix the problem, but customer service isn’t part of the job description for the CentOS developers.

You do get what you pay for. And if you pay nothing, there’s little cause to complain. CentOS 6 will be out, but when is anybody’s guess at this point.

Comments on "Send in the Clones: The Long Wait for CentOS 6"

djflux

Many readers may not be aware of the fact that CentOS is completely community driven, and no one gets paid at all to work on the project. I didn’t know this until I listened to a podcast with one of the project’s contributors. Considering that we have a CentOS at all is a testament to the hard work that all of the contributors put in to making it a superb enterprise Linux OS.

Check out a great interview with Karanbir Singh of the CentOS project on the FLOSS podcast for some more insight on how the project works – http://twit.tv/floss142

Reply
diesel

Why do people feel rushed to ship out a new release inherently buggy software? I say CentOS 5.5 is the greatest, open source enterprise distro available and should be left alone :-)

Reply
    digimer

    except that drivers become an increasing issue, as does software version. For example, the Cluster Stable v3.x is worlds better that Stable 2.x, which is all you get in C5.x.

    Reply
    portablenuke

    RHEL 6 has improvements for when it’s run as a guest VM. That alone has me ready to start deploying it.

    Reply
pjwelsh

You are generally correct in the statement of “it’s not necessarily easy to jump into CentOS <developer> contribution”. And yes the developer group can be a tough bunch of nuts. And yes there are issues that need to be resolved… However, developer volunteering *now* with no previous contribution will not be helpful. It will take more developer time to assist in bringing a newcomer up to speed than be helpful in any way.

I do want to say that the forum http://www.centos.org/modules/newbb/ group is VERY helpful and the solutions to issues are top notch. There are MANY standouts in that group the will attempt to help in every practical way. The forum group *DOES* welcome contribution at all times, also ;)

Reply
mickrussom

They would get a lot more donations at CentOS if they articulated what they needed to get stuff done, and to do that they need at least a rolling whiteboard. I would have given hardware and money via the company I work at but the erratic releases, poor documentation of the process outside the “inner circle” and the total lack of timelines spooks most companies from wanting to donate to this project.

They are screwing themselves by being a closed-opensource project with massive key man issues. (e.g., if a few key guys quit or die its nearly impossible for the community to pickup where they left off)

Horrible leadership-to-community communication skills.

Its too bad, its the best of the RHEL clones and one of the best OSes out there now.

Reply
monteslu

Centos 5.5 is solid. The benefit of releasing a Centos 5.6 seems a whole lot smaller than focusing on 6.0.

RHEL 6 can run things like Chrome and other new apps, while Centos 5.x can’t.

RHEL/Centos is not only used as a server OS.

Reply
    gerdesas

    There are, literally, millions of entrenched C5 installations around the world whereas there are *0* CentOS 6 installs. You provide updates to those systems at risk due to security issues over new release as you have millions of systems potentially at risk. 5.6 having priority over 6 was the only intelligent decision possible; with the advent of the new update to C4 you give that priority over 6 as well.

    Reply
xdigitalvampirex

I actually have been waiting for a CentOS 6.0 release before I even build my home server. Hopefully a Sandy Bridge UP Xeon build. I know they aren’t out yet, but Supermicro already had the boards for them and they should be out this quarter.

As for the “contributor” related comments, ALL software projects or communities need to have better documentation to get new contributors involved. This isn’t exclusive to CentOS, though they may be a shining example. I have this feeling that there would be MANY more contributors to projects if they had better documentation, but it can be a hassle (and boring) to produce. It can take weeks (spare time, of course) to start helping with some projects, and small things like a simple “code layout document” or quick “how-to videos” could significantly cut this time down. On the other hand, perhaps a larger barrier to entry is a good thing. That my be the reason there are so many “fart” applications for the iphone or Android. :) Just a thought.

Reply

Webkit pretty much requires 6.x and its dependencies are so bad, that it’s too difficult to beef up a 5.x install to make it work.

Reply

i want the magazine of linux…
pls send me this magazine

Reply

You made various good points there. I did a search on the topic and found a good number of persons will agree with your blog.

Reply

whoah this weblog is great i like studying your articles. Stay up the good work! You already know, a lot of individuals are searching around for this information, you can aid them greatly.

Reply

Leave a Reply

Your email address will not be published. Required fields are marked *

*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>