Submitted by Though Red Hat is poised to roll out its much-anticipated Red Hat Enterprise Linux 5 release shortly, work still continues on previous versions of RHEL. This week Red Hat rolled out a beta release of its fifth update to RHEL 4 officially tagged Red Hat Enterprise Linux 4.5, providing users with a small taste of the virtualization that is to come in RHEL 5.
what happened to rhel 5 being released on feb 28th then?
now this is a beta of a previous version, hints to me that rhel5 is months off still, and judging by the state of beta2, for good reason.
the same happens with fedora, they’re going to be over a month later than their original f7 deadline.
why don’t they either make acheivable deadlines, or throw more resources in to get things done on time?
Perhaps because they need stable virtualization?
Their system is very stable by now (Enterprise, not Fedora) so, why should they release something that could put in risk their reputation of producing a very stable OS? You know, their main target is corporate environment anyway and there stability and security are #1 and #2 concerns (perhaps #2 and #1 respectively). I think it will be ready when virtualization gets really stable, not before.
Edited 2007-03-03 23:35
> Their system is very stable by now (Enterprise, not Fedora)
RHEL : stable AND stable API (source and binary) : RHEL 4 use Linux 2.6.9 only
Fedora : stable AND no stable API : FC6 launch with linux 2.6.18, last update has Linux 2.6.19, Linux 2.6.20 will be in FC6.
Fedora is not as stable as RHEL. But as a blinding edge system, Fedora is very stable.
“””
But as a blinding edge system, Fedora is very stable.
“””
Yeah. That’s why they find it necessary to release an average of 27+ MB of patches per day on Core alone. Never mind extras. That’s 10 patches per day.
If you add up the updates for FC6 [dc(1) is your friend] and divide by the 131 days since the release, that’s what you get.
Leaving out weekends, when patches are not normally issued, that works out to 38MB and 14 patches per release day.
In fact, 4.5 months after the FC6 release, Fedora has issued 1328 patches representing 3.5 *gigabytes*. And remember, that’s just core, not extras.
Sounds really stable to me.
Not even Rahul can defend these numbers.
I predict that someone is now going to try to convince me that it’s because FC is so great about issuing security patches, and that 38MB of patches per day is really a good thing. They’ll ask me if I would prefer that FC *not* issue these necessary patches. Just watch.
Edited 2007-03-04 16:45
“In fact, 4.5 months after the FC6 release, Fedora has issued 1328 patches representing 3.5 *gigabytes*. And remember, that’s just core, not extras. Sounds really stable to me. Not even Rahul can defend these numbers”
Cute. I don’t need to defend the numbers but I can very well clarify and debate the interpretation of these numbers.
Stability generally can mean two very different things to people.
In the first case, it is a measure of robustness which Fedora does strive to be good at it though far from being perfect.
The second measure is ABI stability and general level of conserativeness. Since usually upstream open source projects dont care much about that (with notable exceptions like GTK), individual distributions like RHEL do heavy backports to older versions for retaining ABI stability as much as possible.
Fedora in contrast tends to stay closer to upstream versions. Historically about 10% of updates are backports in Fedora Core. Maintainers in Fedora coordinate and push out updates for packages in the repository simultaneously for ABI breakages and are less strignent compared to RHEL about it especially for things like kernel.
Also while distributions like RHEL usually tend to provide only security and bug fixes (with few exceptions for new hardware support or some features like say Xen in RHEL 4.5), Fedora updates includes new features too. If upstream releases a newer versions, they get to Fedora development and then to Fedora general releases fairly quickly while RHEL is much more conservative. There are advantages and disadvantages to both approach. The differing policies are aimed at satisfying different kind of end users.
“””The differing policies are aimed at satisfying different kind of end users.”””
Yes. Fedora is aimed at users who don’t mind waking up in the morning to a broken machine, but want cutting edge.
My argument is not with people who claim that Fedora is a good cutting edge distro, but with people who want to claim that it is reliable, as well.
I have much experience with it, CentOS, and Ubuntu, in similar environments, and can say without a doubt that Fedora stands out as being less reliable than the other two on a day to day basis… which is obviously fine for some users and not fine for others.
Of course, one requirement for Fedora is a fast broadband connection. I installed FC6 on a machine the other day, and it took all night to download and apply the 1.2GB of patches that the machine required, over its 768kbit/s fractional T1 connection. It would have taken 3+ days of continuous downloading on a dialup connection.
On the other hand, I installed another distro, which was released about the same time as FC6, last night and needed “only” 200MB of patches. Still a lot. But only 1/6th as much as FC.
Edited 2007-03-04 18:06
> which is obviously fine for some users and not fine for others.
Hummmmm, yes.
RHEL != Fedora != Ubuntu != Mandriva etc…
Everyone have its own best distribution.
“””
Hummmmm, yes.
RHEL != Fedora != Ubuntu != Mandriva etc…
Everyone have its own best distribution.
“””
Absolutely. And promoting a distro as something that it is not does a disservice to *everyone*, including, and even *especially* the distro being misrepresented.
Because then people try it out expecting it to be one thing, and are disappointed that their expectations were not met.
When I call someone on public statements they make about Fedora, that is almost always my point. But I think a lot of people miss it, thinking that I just want to trash the distro.
-Steve
“””
“In fact, 4.5 months after the FC6 release, Fedora has issued 1328 patches representing 3.5 *gigabytes*. And remember, that’s just core, not extras. Sounds really stable to me. Not even Rahul can defend these numbers”
Cute. I don’t need to defend the numbers but I can very well clarify and debate the interpretation of these numbers.
Stability generally can mean two very different things to people.
“””
BTW, since we’re kinda sorta on the topic… any chance of Fedora Unity’s respins becoming available for network installs (as opposed to downloading the entire set of CD images)?
This would at least make FC network installations tolerable. Or for that matter, as freewheeling as FC updates are anyway, why not just have the installer include the updates repository during installation so that it installs the updated packages rather than the original packages during the initial install process?
” BTW, since we’re kinda sorta on the topic… any chance of Fedora Unity’s respins becoming available for network installs (as opposed to downloading the entire set of CD images)? ”
Dont know. They are a different group. Log in to freenode #fedora-unity and ask. They probably use just torrents because of bandwidth limitations. Pungi which both Fedora and unity folks currently use can produce whatever combinations of packages you want.
https://hosted.fedoraproject.org/projects/pungi
“Or for that matter, as freewheeling as FC updates are anyway, why not just have the installer include the updates repository during installation so that it installs the updated packages rather than the original packages during the initial install process?”
This is already possible in FC6 using kickstart. You can specify the updates repo in the interactive installation in Fedora too but there are some wrinkles that needs to be sorted out and it is not activated by default (probably not for Fedora 7 either though Fedora 8 is possible) . FC6 also provides repository metadata that classifies bugs as bug fixes, security fixes etc and one of the developers is planning on a yum security fixes only plugin which might be useful if you dont want to get all updates.
Criticism on robustness is valid ( but it is better to help rather than spread misconceptions and describe implementation flaws as design) but due to Fedora updates policy of including newer upstream versions generally, the number of updates are not a indication of security fixes or robustness. There are other indicators like regressions etc. It is getting better. Sooner if you help.
“””
Criticism on robustness is valid ( but it is better to help rather than spread misconceptions and describe implementation flaws as design) but due to Fedora updates policy of including newer upstream versions generally, the number of updates are not a indication of security fixes or robustness. There are other indicators like regressions etc. It is getting better. Sooner if you help.
“””
I’m not spreading misconceptions. I’m commenting upon what has been true for years now. An implementation detail that remains true for years may as well be design.
As to “sooner if I help”, we’re not talking about a few guys trying to make their own distro and looking for help. Fedora is one of the best funded distros out there. Quite possibly *the* best funded. Most of its many developers are paid by Redhat. If RedHat wanted it stable it would be stable.
Why should regular folks contribute their time, for free, to try to make it stable if RedHat doesn’t care to make it stable, and arguably does not want it stable? Why should they not contribute their time and effort to a distro for which stability is a major goal rather than a secondary consideration?
”
I’m not spreading misconceptions. I’m commenting upon what has been true for years now. An implementation detail that remains true for years may as well be design”
A implementation detail regardless of the length of its time should always be treated separate from design. Software engineeeing 101.
“Why should regular folks contribute their time, for free, to try to make it stable if RedHat doesn’t care to make it stable, and arguably does not want it stable? Why should they not contribute their time and effort to a distro for which stability is a major goal rather than a secondary consideration?”
Fedora is a community project where everyone can contribute. If a single entity has to take on all its tasks then it would purely be a commercial product. What makes it a community project where everyone can contribute and does to various extends is the volunteer contributors. Much of the packages you can get from Fedora Extras, L10N, documentation etc.
Red Hat does employ people to lead the QA team in Fedora and write automated tests that help improve robustness. So there is no doubt that there is a focus in it. RHEL 5 for example is derivative of FC6 so there is a commercial interest in doing this.
However the scope of Fedora is much larger than RHEL and including the latest software makes it much more difficult to provide the level of robustness and hence more community participation is necessary.
http://fedoraproject.org/wiki/FedoraTesting
Fedora is a complete leading Linux distribution that users can get. Users pay back by their contributions. Simple Free software mechanism.
Red Hat does not sell RHEL box.
Red Hat sell support. Part of the job support is to provide erratas (bugfix, security), features (like RHEL 4.5), and also new system that bring new features but break API. If you have a RHEL subscription you can upgrade (or downgrade :-)) to any RHEL release available.
RedHat’s Customers don’t pay to repair crapy system. They pay to get a rock solid system.
Edited 2007-03-04 00:05
A RHEL 5 Release Candidate has been out for a bit. Release Candidate is not for general consumption. The last public release was Beta 2, which was released last November.
RHN Satellite was updated on Feb 28th to handle the changes in RHEL 5.
https://www.redhat.com/archives/rhelv5-beta-list/2007-February/msg00…
Last word is RHEL 5 will be released in March.
http://news.com.com/2061-10795_3-6159272.html?part=rss&tag=2547-1_3…
> Last word is RHEL 5 will be released in March.
Jay Turner knows the release date :
http://www.redhat.com/archives/rhelv5-beta-list/2007-February/msg00…
The release of 4.5 doesn’t really mean anything special. Redhat releases updates to all major versions periodically. For RHEL4, we’ve seen 4.0, 4.1, 4.2, 4.3, and 4.4. It’s about time for a 4.5. Adding virtualization to the kernel in 4.5 is a bit exceptional, however.
RedHat tries not to set hard deadlines. However, they do advertise a release schedule for RHEL of every 18-24 months. So far, they have released closer to every 18 months; I was expecting RHEL last September or so.
But they are currently at only 24.5 months between RHEL4 and RHEL5.
I’m as impatient as anyone. But they are only a couple of weeks past their stated policy.
Compare this to Debian, which got a lot of flak for taking *3.5 YEARS* between releases last time, and vowed to do an 18 month release this time, due… December of last year. They are now 3 months over the dealine with no release in sight.
If you are going to go after projects for not sticking to deadlines, they’d be a much better target.
I’m going to go out on a limb and predict that due to unexpected problems and conflicts, we will not see a new Debian release this year.
Yeah, I know that’s not the general opinion on that matter, but I want to get my prediction registered somewhere, in writing, early, so I can point back to it next year. 😉
The release of 4.5 doesn’t really mean anything special. […] Adding virtualization to the kernel in 4.5 is a bit exceptional, however.
According to TFA, the virtualization feature added in RHEL 4.5 was just a new (and optional) kernel so that customers can virtualize RHEL4 as a guest on RHEL5 as soon as the latter comes out. This is to make sure that customers have a smooth and easy migration path to RHEL5.
…Unlike how it works with that other platform vendor.
I’m not so sure that Debian will fail to release Etch in 2007, but I’m with you in principle. The thing I notice about criticizing the Debian Project is some people will fight you tooth and nail, but you’ll get modded up for being right… 😉
“””The thing I notice about criticizing the Debian Project is some people will fight you tooth and nail, but you’ll get modded up for being right… ;-)”””
I believe that things have gotten much better over the last couple of years.
It used to be that making constructive criticisms of Debian was more trouble than it was worth. The hordes would descend upon the well meaning critic with epiphets and personal attacks upon his own person, his spouse, and his immediate relatives. (Even family pets were not immune to attack.)
I believe that the proliferation, and in one case, wild success, of Debian derivatives which improve upon Debian proper, have demonstrated to many that the developement of Debian itself is not guided by the Divine Hand of God, and that Stable, Testing, and Unstable are not the Holy Trinity that many once thought they were.
Ironically, the success of Debian derivatives has embittered a few Debian developers, but the situation is still far better than it was only 3 years ago.
I cant wait for all the new features in RHEL 5. fedora 7 looks awesome too. Fedora gets a bad wrap sometimes, but their stuff is way out on the cutting edge usually. You want safe, use RHEL. ME, I like to try the new bells and whistles. And even if I have to reinstall, its still a hell of a lot faster than Windows to get back to a patched and stable state.
I’ll be waiting for the equivalent Centos release though..
Too bad, Centos is waiting for RHEL.
Waiting for Centor is waiting for RHEL 🙂
If RH were to release RHEL 5 this month as they said, I doubt they would have unleashed RHEL 4.5. This release looks like some meat for the impatient user base while they complete RHEL 5, with this move they can legitimately ask for more time, maybe a few months.
that doesn’t make any sense at all. RHEL4 will be supported and get updates for several years to come. It’s update timeline has absolutely nothing to do with RHEL5. Funny thing is, is that the people I have seen screaming loudest over RHEL5 being a bit late, are the same people that are very unlikely to be using it.
Funny thing is, is that the people I have seen screaming loudest over RHEL5 being a bit late, are the same people that are very unlikely to be using it.
Why? If you use CentOS, you don’t use RHEL but you still can whine at RH.
Man, I wish I could mod that comment up even more.
It’s probably normal to feel a little frustrated if you’re waiting on some new functionality, but I can not imagine many production environments that would choose a new, potentially flaky feature over a stable release.
RHEL is built to run on production systems. If you want bleeding edge, then look no further than the dozens of prominent distros that will get you there. (And don’t be surprised when things start to break.)