Fedora Legacy Shutting Down

An anonymous reader writes to pass on the news that the Fedora Legacy project is going away. The project has been providing security updates and critical bugfixes to end-of-life Red Hat and Fedora Core releases. From the article: "In case any of you are not aware, the Fedora Legacy project is in the process of shutting down. The current model for supporting maintenance distributions is being re-examined. In the meantime, we are unable to extend support to older Fedora Core releases as we had planned. As of now, Fedora Core 4 and earlier distributions are no longer being maintained. Discussions... on the #Fedora-Legacy channel have brought to light the fact that certain Fedora Legacy properties (servers) may be going away soon, such as the repository at http://download.fedoralegacy.org and the build server."

Comments

RH pushing EL

[Ice Wewe]

IMHO, dropping support for a system that is only 3 years old is a mistake. Everyone who hasn't upgraded to the latest versions of Fedora will loose support. Ubuntu provides 5 years of support with most of their major releases, so would it be so hard for Redhat to follow? I think that Redhat is trying to push it's Enterprise Linux a bit too strongly. I'd like to see the support window for future releases of Fedora to be extended.

Re:RH pushing EL

[Wdomburg]

Erm, Ubuntu supports for 18 months for most of their releases. Only one release has been designated "long term support" and that's only 5 years for servers; the desktop version is only supported for 3 years.

And, on top of that, Fedora Legacy is not Red Hat, is not affiliated with Red Hat, and is not sponsered by Red Hat. As such their actions don't reflect on Red Hat.

Re:RH pushing EL

[arth1]

IMHO, dropping support for a system that is only 3 years old is a mistake.

I hate to say so, but Microsoft usually does better than this. I still get security updates for Windows 2000, even though the last service pack came out in 2003. With Gentoo, packages drop out sometimes as little as 3 years after they appear, which is not good, and SuSE isn't any better. I can understand dropping full support, but security support should still be there. After all, some customers can't easily upgrade because of legacy[1] software. This is particularly true for those who for some reason need to stay at kernel 2.4 or below. There's not a lot of options for support of that anymore.

[1]: The definition of legacy is "something that works".

Regards,
--
*Art

Re:RH pushing EL

[LnxAddct]

What is up with everyone pointing fingers at Red Hat? Fedora was never supported by Red Hat. Fedora is a community driven project and the Fedora Legacy project just didn't have enough community interest. Now the Fedora community is rethinking how it can support older releases... Red Hat has nothing to do with any of it.
Regards,
Steve

Re:RH pushing EL

[Znork]

Compare Fedora with the Ubuntu minor releases and RHEL (or CentOS) for the major LTS releases.

If you look at jkeatings blog is sounds more like an extension of support time for the FC releases is likely, and those requiring long time support are pointed towards RHEL/CentOS, which probably are better matches for long term needs anyway.

Personally, I'd say the existence of CentOS made the legacy project far less essential.

Re:RH pushing EL

[Vellmont]

I still get security updates for Windows 2000, even though the last service pack came out in 2003.

Right, because Microsoft has a policy to support a release for something like 8-9 years, and they're a company that's supported by software sales. Windows 2000 is also a production level OS. What was the policy on Fedora? 2 releases? How much did you pay for it? $0? Also, FC has always been understood to be a bleeding edge system that you shouldn't put any kind of production system on.

The point is if you want support for an OS for many years you should choose one that has a long term support policy. Redhat Enterprise Linux supports a release for 7 years. Ubuntu has a release that has 5 years of support. If that's what you want, choose one of those two distributions. But don't complain when a 3rd party support organization with little or no income decides to stop supporting a distribution that's never intended to have long term support.

Re:RH pushing EL

[asuffield]

Ubuntu provides 5 years of support

That's a popular myth, but Ubuntu has not existed for 5 years. A more correct statement would be "Ubuntu states an intention to provide 5 years of support". That's not the same thing as a real support contract from a company that you're sure will still be around in 5 years time. Anybody could state an intention to provide N years of support, but that doesn't mean that you'll actually get it. (Redhat - not Fedora - have been providing such long-term support contracts to certain corporate customers, although I'm not sure what their longest duration one is. They just don't provide them to everybody who happens to buy an RHEL license)

Re:RH pushing EL

[metalhed77]

Even though it's only one release, it's a big deal. I'm using Ubuntu 6.06 Long Term Support in a server I recently built my company for just that reason. Having the peace of mind that there will be regular backports for 5 years on my free OS of choice was a deal breaker.

Re:RH pushing EL

[rrohbeck]

I've been a faithful RedHat/Fedora Core user for years, but obsolescence happens too fast for anything but test or playground machines these days, where you don't mind a complete reinstall every now and then.
Since I don't need RedHat's support, my servers now run CentOS, and everyhingthing else is on Ubuntu.

Re:RH pushing EL

[Wdomburg]

Fedora != Fedora Legacy

Re:RH pushing EL

[Wdomburg]

I've been running Ubuntu on my laptop for a while now (a few days ater 6.06). Seems a solid enough distribution, so far, but I'd personally hold off using it on a server until they a longer track record and a formal definition of "support".

When you're talking server platforms, Fedora was never a viable platform (despite some folks insistance on running it as such). For something Red Hat flavoured, you'd want to look at CentOS.

Re:RH pushing EL

[Master of Transhuman]

I'll point out the obvious - you haven't received five years of backports - yet.

Tell us again in five years when Ubuntu has kept their promise.

I'm not saying they won't - but five years is a long time in IT. Right now, you can't guarantee Ubuntu will even exist in five years. Red Hat has a track record - of that long at least.

Also, I find this notion that servers can't be touched for five years because "if it ain't broke, don't fix it" is reaching for the moon in terms of reliability - and strikes me as more a product of laziness than competent system administration. You're going to have worse problems than that over five years, one way or the other, so it's really an irrelevant issue. All you really want - and need - is not to have the servers crashing every six or twelve months when a new release comes out. That's what test servers are for.

Re:RH pushing EL

[metalhed77]

You're right about it not being guaranteed, and if I have to upgrade it I will. However, the less time I have to spend upgrading servers the more time I have for stuff my bosses actually care about.

Re:RH pushing EL

[petermgreen]

Having the peace of mind that there will be regular backports for 5 years on my free OS of choice was a deal breaker.
don't you think its a little premature to assume that? ubuntus first release was just over two years and has made one LTS release which they have been supporting for less than a year.

When dapper+8 (4 years after dapper) is released and both dapper and dapper+4 support have been provided for all that time then assuming ubuntu looks healthy it will be time to rely on ubuntus long term support policy.

finally remember the LTS policy and even the 18 month policy for normal releases only apply to stuff in main so if you use stuff from universe you either have to upgrade every 6 months or track security issues on the universe stuff yourself.

Re:RH pushing EL

[Wdomburg]

Five years is hardly an unreasonable life cycle in large scale IT deployments. It's not laziness to avoid massive expenses - the upgrade itself, retooling config files (e.g. Apache 1.x to 2.x), recertifying hardware, porting or rebuilding software, running everything through a new QA cycle, etc. There's also issues of SLAs which may require customer approval for major software changes or incur penalties if critical systems need to be downed for the upgrade (and god forbid anything goes wrong).

Edge into even a medium-sized company and an upgrade may cost man-years of labor from several departments. You damn well better have a solid business case before you recommend an upgrade.

Re:RH pushing EL

[Master of Transhuman]

SLAs are an external matter.

As for the rest, my guess is that in most cases with any medium or large business, business needs (new services, new marketing, new regulatory, whatever) are going to force doing all that anyway, regardless of what the OS vendor actually does.

I'd like to see a survey of companies who will admit to having run the exact same server for five years (recent years, not ten years ago) with no changes.

Most of the stuff you mention is either something that shouldn't take "man years" of effort, or, in the case of porting software and QA, depends on factors outside of the OS itself and are even more likely to require changes over a five-year period in today's environments.

Finally, all of this stuff is a matter of IT management PLANNING. If your planning is decent, an OS upgrade should not be a make or break event. Waiting until the last minute and doing it when you HAVE to do it is how you get "man years" of effort involved.

Just as the best way to maintain a car is to know hoe many miles each part is certified for, then replace it BEFORE it breaks, the best way to maintain a server is scheduled upgrades - not trying to run it into the ground for five or more years.

THAT is lazy - and incompetent.

Re:RH pushing EL

[Wdomburg]

As for the rest, my guess is that in most cases with any medium or large business, business needs (new services, new marketing, new regulatory, whatever) are going to force doing all that anyway, regardless of what the OS vendor actually does.

That's largely a matter of software running on an operating system, not the operating system itself. With commercial software you're actually more likely to have a requirement on an older platform than a new one (plenty of apps out there are certified for nothing newer than RHEL3).

And really server software doesn't evolve all that quickly. Apache 1.x can deliver fancy shmancy "web 2.0" pages just dandy, with MySQL 4.x as a backend even.

Most of the stuff you mention is either something that shouldn't take "man years" of effort, or, in the case of porting software and QA, depends on factors outside of the OS itself and are even more likely to require changes over a five-year period in today's environments.

It should take as long as it takes. And the reality is that it is man-years of effort to do right. Upgrades in a production environment have to be heavily planned and handled with a lot of care. Regardless whether porting and QA are part of the upgrade itself, they are an added cost incurred by an upgrade cycle.

Finally, all of this stuff is a matter of IT management PLANNING. If your planning is decent, an OS upgrade should not be a make or break event. Waiting until the last minute and doing it when you HAVE to do it is how you get "man years" of effort involved.

Who said anything about waiting until the last minute? An overly aggressive upgrade cycle ensures insufficient planning. A five year upgrade cycle affords you the opportunity to start evaluation and planning early, have a leisurely pilot and test cycle, and leave plenty of wiggle room in the rollout to make sure you hit the target completion date.

Just as the best way to maintain a car is to know hoe many miles each part is certified for, then replace it BEFORE it breaks, the best way to maintain a server is scheduled upgrades - not trying to run it into the ground for five or more years.

My "car" is certified for seven years and gets regular scheduled upgrades. What you're saying is I should trade in a well-maintained vehicle that runs beautifully just because there are newer shinier carson the lot.

Re:RH pushing EL

[Dan Ost]

When a package gets dropped from portage, it's because nobody is willing (or able) to maintain the ebuild (and any required patches). That only happens with unknown software that nobody uses or software that has been completely outdone by its competition such that nobody uses it anymore.

I'm just going out on a limb here, but if your business chose to use certain software, you probably didn't choose software that nobody else uses. That right there will probably prevent it from being dropped from portage (if it's important to more than one person/organization, it's usually well supported). In a worst case scenario, you could simply maintain your own ebuild and have it included in portage (if it works, it's no problem to get it officially into portage, although they might ask you to be the official maintainer).

Re:RH pushing EL

[Master of Transhuman]

I didn't say replace the CAR, I said replace the PARTS!

That's HOW you make a car run seven - or ten or fifteen - years.

It's the same with a server. You're suggesting running the car for ten years without changing a part just to avoid putting in a part that won't work right. While software isn't as simple as a car part, this is still incorrect.

And it doesn't take five years to plan a production environment upgrade - that's ridiculous. The PLANNING can be done in a month - it's the actual upgrade that can take a while if done properly over time. And even a test cycle could be done in a relatively short time.

We're also not talking about server software not evolving quickly. In that situation, you don't have to upgrade the server software itself. We're talking about the OS needing upgrades due to lack of support - which is part of the business of software. Software cannot be supported forever or even the number of years desired by businesses using it. That's a simple fact. The manpower on the part of the software producer isn't there to do that for the number of customers using it. While you can probably easily find a car mechanic who can support a car built ten years ago, it isn't that easy to find the equivalent in software due to the complexity of software and the economics of supporting the few people using it. A software company is not a single mechanic.

There might be room for a company which does nothing BUT support old software, but it wouldn't be a development company, it would be a maintenance company and the economics would be different.

It would be nice if software were as modularized and documented and controllable as a car part, but it isn't. Business needs to accept that fact and use scheduled upgrades to deal with it. The upgrades might be three or five years, or less or more, but it has to be done. The difference between Microsoft Windows and Linux distros is that it's possible to control WHEN the upgrade is done to a greater extent with Linux distros.

"If it ain't broke, don't fix it" is primarily a combination of laziness and an indifference to opportunities costs.

Re:RH pushing EL

[arth1]

The problem with Gentoo is that often, older versions are removed from portage, presumably because the maintainer only wants to maintain the new version. This bites those who can not upgrade to the newer version. A prime example is dev-lang/perl, where you now have the choice between perl-5.8.8-r2.
That leaves those who have to stay at perl 5.6 (or even earlier) kind of out in the cold. And 5.6 was retired as early as Feb 2004, which is *WAY* too early.
There's large vendors that still ship their product with a perl requirement older than what gentoo supports (like servers from MCOM/Netscape/iPlanet/SunONE/SunJava or whatever they call themselves today). It truly sucks to have to maintain a non-OS supported version in addition to the OS one, and make sure there's no @INC pollution between the two.
Similar with other ebuilds like BIND (I know people who have to stay at BIND 8 due to zone files being propagated to other companies' servers that won't talk to BIND 9).
Or sendmail, exim, or any of a number of other major packages.

Is 5-6 years support for major packages like these asking too much? Or at least more than a single supported version/revision/subrevision, like there is for perl?

As long as a new version of a major package is not 100% API compatible with the older versions, the older versions should not be removed if it's only a few years old, no matter how tempting it is to support only the latest. Cause people will depend on them. The way they're being yanked out these days, I can't honestly recommend gentoo for corporate or business use.

Re:RH pushing EL

[Dan Ost]

Ebuilds for older versions of everything in the portage tree are available in CVS if you need them. I do realize that at some point the system changes enough that old ebuilds might not work properly without tweaking, so if you have vendor supplied software and can't build it yourself, perhaps Gentoo isn't a good match for it.

Re:RH pushing EL

[Wdomburg]

It's the same with a server. You're suggesting running the car for ten years without changing a part just to avoid putting in a part that won't work right. While software isn't as simple as a car part, this is still incorrect.

Most operating system vendors put out these neat things called patches (or updates, or errata) to fix things that are broken. Those are parts. Moving to a new version of the entire operating system is akin to buying the next model year.

And it doesn't take five years to plan a production environment upgrade - that's ridiculous.

I never said it did.

The PLANNING can be done in a month - it's the actual upgrade that can take a while if done properly over time. And even a test cycle could be done in a relatively short time.

That depends entirely on the company and their business requirements. If you've got a relatively large number of servers, a wide range of applications or constrictive SLAs your test cycle will be considerable. There's also the question of how much spare manpower you have for the job. Having to bring in consultants to oversee regression tests because your schedule is compressed and/or your normal QA people are otherwise occupied is added overhead that can usually be avoided.

We're also not talking about server software not evolving quickly. In that situation, you don't have to upgrade the server software itself. We're talking about the OS needing upgrades due to lack of support - which is part of the business of software. Software cannot be supported forever or even the number of years desired by businesses using it.

It can't? Better tell that to Red Hat (7yr support cycle), SuSE (7yr support cycle), Sun (12yr support cycle), IBM (5-6yr support cycle), Hewlett Packard (10yr support cycle), Microsoft (10yr support cycle), etc.

It would be nice if software were as modularized and documented and controllable as a car part, but it isn't. Business needs to accept that fact and use scheduled upgrades to deal with it.

The important bits for most server applications - web servers, database servers, file servers (in the case of *nix at least), etc are modular.

The upgrades might be three or five years, or less or more, but it has to be done.

I never said anything about an indefinite upgrade cycle. You said "this notion that servers can't be touched for five years ... is reaching for the moon in terms of reliability" and I disagreed with "five years is hardly an unreasonable life cycle in large scale IT deployments.

I stand by my statement. Five years is a fairly common lifecycle. Three years is possibly more common for actual deployment (ties nicely to common lease terms) but even then you want the support cycle of the platform you're installing to be longer to allow for testing prior to deployment.

Re:RH pushing EL

[turbidostato]

"IMHO, dropping support for a system that is only 3 years old is a mistake."

IMHO, the mistake is using a distribution that won't guarantee its long-time stability or upgrading procedures for any production-grade purpouse.

"Everyone who hasn't upgraded to the latest versions of Fedora will loose support."

As if they wouldn't have to know that would happen when they firstly started to use Fedora.

"Ubuntu provides 5 years of support with most of their major releases"

Ubuntu provides 5 years support on *one* of their major releases and since only more or less one year has passed for that one only distribution it is still to be seen if they will be able to stay after their promises.

"so would it be so hard for Redhat to follow?"

Humm... Red Hat already has a long lived product line, and being older that Ubuntu they eventually can claim that they have already demonstrated they can do it, while that's not the case with Ubuntu.

I'd like to see the support window for future releases of Fedora to be extended."

That would be an interesting move:

RH CTO: Hey! why don't we expend a lot of money supporting a free (as in free bear) product that is in direct competition with our number one money maker product line so Ice Wewe is happier?
RH CFO: Jesus Christ! how is it possible we didn't think of it earlier? Of course yes!

Re:RH pushing EL

[turbidostato]

"SLAs are an external matter."

To the IT department. Many companies have internal SLAs from IT towards other departments. Anyway SLAs are for a reason which is still quite valid within the company: don't disrupt bussiness.

"I'd like to see a survey of companies who will admit to having run the exact same server for five years"

I'd say you would be surprised with the answer. They are high numbers.

"If your planning is decent, an OS upgrade should not be a make or break event."

Still, planning costs money, and executing plans costs even more money.

"Just as the best way to maintain a car is to know hoe many miles each part is certified for, then replace it BEFORE it breaks"

The stupid car analogy. Just for the record, software parts don't grow old nor break apart for excesive usage. As long as your needs don't (localy) change the software just continues humming fine.

"the best way to maintain a server is scheduled upgrades "

I'd say "planned" better than "scheduled" but still, planning for stability costs less than planing beforehand for unneeded change ratios. If my company changes their objectives or needs so I have to rebuild our IT structures from the ground up, so be it: that's my company decision, it's not Red Hat's or SuSE's or Microsoft's one. And I can schedule (ie) five years turnover on Ubuntu, three on Red Hat or six months on Fedora, just to be in the exactly the same position. Not taking this fact into account it's not lazy, but it *is* stupid and incompetent.

Debian supremacy!!

[Anonymous Coward]

This is just another little step to Debian supremacy!!! Go GNU/Linux Debian!!! The Universal Operative System!!!

Re:Debian supremacy!!

[Eto_Demerzel79]

The next step to Debian supremacy is to get an installer that will save Linux rookies from having to edit scripts or for that matter use a terminal, which would include automatically installing synaptic and let them choose any additional software they like via KDE/Gnome. Also, an effort should be made to make Wine available so that they can use some Windows apps if necessary. I've been using Debian since 1999/2000 and I would still rather reboot my box into Windows than configure Wine to run some app. Never underestimate how lazy and impatient some people can be. I have installed Debian for many new Linux users and they either ask me for something easier (and I give them FC) or they just leave the Linux partition alone and continue using Windows. my $0.0002

Re:Justification?

[93 Escort Wagon]

You mean, how are Linux fans going to justify Red Hat not supporting the older versions of software that doesn't cost anything? Technically Red Hat doesn't even support the current version of Fedora Core. If you want support for multiple years, buy one of the flavors of Red Hat Enterprise.

Are you going to complain next that Microsoft isn't supporting Vista's beta 2 anymore? It's pretty much the same thing.

To be expected...

[Junta]

Fedora's whole mission is pretty much implement the cutting edge and let people experiment and play with it. The target audience was inteded to be those desktop users/enthusiasts who would jump on the current release anyway, since it is free. If any business saw a distribution like Fedora and thought it a good idea to base their infrastructure on a Fedora Core release, they are now getting basically what they asked for.

I don't think this is necessarily a bad thing, it's more an acceptance of what people should have perceived all along, Fedora isn't trying to be that kind of distribution. If you want that kind, go to a commercial vendor (RHAT, Novell) or go with something like CentOS or Debian, which have clear missions/policies that align with that sort of usage. Could consider Ubuntu, but I wouldn't be that confident in Canonical yet, but the LTS is at least a stated mission and Canonical has a vested business interest in being considered a serious business worthy option, while Fedora obviously hs no such vested interest. Similarly, I wouldn't use Gentoo or OpenSuSE in those contexts either, their missions are valid, but not in line with common business desires/needs. Debian and CentOS do, and generally end up 'boring' in the eyes of enthusiasts, and Fedora Core, Gentoo, Ubuntu's 6 month releases, etc all are generally more interesting to the enthusiast, but can't provide legacy support and the cutting edge all the time.

Re:To be expected...

[PybusJ]

True, and yet I'm not sure that this is the best decision for Fedora/RedHat.

I've been using RedHat for over 10 years -- since RH3.03. Initially, I ran downloaded RH on my workstations and bought a licenced copy for any servers I deployed. After RH9, I ran Fedora on my workstations and either RHEL or one of the respins on servers. The Fedora Legacy project provided valuble support for production RH7/RH9 servers in the changeover period (somewhat badly handled by RHAT), and extended support for Fedore Core releases by a valuable few months.

In recent months it's become obvious that Fedora Legacy wasn't functioning any more leaving FC with just 12 months support. With a 6 month release rate this means that when I come to install a machine I have on average 9 months of updates left. This just feels too short. I choose Fedora over RHEL or clones on desktop/laptop because it supports more hardware, uses more uptodate components, and more (OSS) 3rd party software comes with FC downloads/support. I can understand Redhat wanting to keep some clear water between their Fedora project and Enterprise products, but 12 month support (or the recently suggested 13) is too little. I use my machines to get things done, not just as a toy, and don't wish to spend time on upgrades for at least a year after installation.

In the last two months I've started install Ubuntu on new machines and switching instead of upgrading old. The amount of available software and community support are comparible, it's just meant getting used to a few Debianisms instead of the RHisms I've been used to. My servers currently still run RH family OSs, but as I get more familiar with Debian like systems and do development on Ubuntu systems, new services will naturally be deployed to Debain/Ubuntu rather than RH.

I wish no ill will to Fedora, but it no longer meets my needs. We'll have to see about Canonical/Ubuntu long term but their stated goals with LTS versions sound ideal. In retrospect I wish I'd contributed to the Legacy project but time as ever was too pressured.

Re:RHAT just said fsck you to those who do not pay

[Wdomburg]

RHT said nothing. Fedora Legacy, "a community-supported open source project" which "is not a supported project of Red Hat, Inc." did.

Microsoft must feel vindicted

[bogaboga]

Call me a troll if you may but, Microsoft must be laughing on hearing such news. You see, there were [Linux] zealots here saying Microsoft should release versions of Windows it does not support to the "community." This "community" could do a better job supporting these operating systems.

My question though is whether what is happening to this Fedora Legacy would not happen to released Windows versions. I have my doubts.

Re:Microsoft must feel vindicted

[phoenix.bam!]

Since Fedora is free there is little reason to still use old releases. Upgrading is free. Windows, not so much. I am running Fedora Core 6 right now and I don't see why you'd want to run an older release. Stable and the software is up to date. I wouldn't trust it on a server but that isn't what it's designed for.

Re:Microsoft must feel vindicted

[Vellmont]

Microsoft must be laughing on hearing such news.

I hardly think Microsoft really cares. As Bill Gates pointed out, Linux is like the multi-headed hydra. It doesn't really matter of one of the heads is cut off. Anyway, FC had never been about long term support. If you chose it expecting that you'd be able to run the same OS release for years, that's your mistake.

This "community" could do a better job supporting these operating systems.

I don't think anyone has said that there's a need to support every OS Microsoft has come out with. I kind of doubt there's many people that want to support DOS 3.0 or Windows 1.0 for instance. Windows 98 might still have a following though. Community support relies on community interest. Try to remember that all distributions of Linux aren't the same in terms of needs, just like all versions of Windows aren't the same in terms of needs.

Open-source software doesn't guarantee support for software for forever, it just makes it possible when there's enough need for it. I'll guarantee you that if there's a big need for someone to support old releases of FC (and that need outweighs people just changing to a different distribution), some people will get together and support it.

I'll give you an example. Currently I have a server running Redhat 9. It runs just great and I've no reason to upgrade to something newer. Support from Fedora Legacy has recently ended. Would I LIKE to keep the machine running on RH9? Sure... why not? But most everyone else has long since moved on. So far I haven't heard of anyone getting together to offer support for RH9, so I'll likely have to spend an afternoon upgrading to Centos 4/5.

Re:Microsoft must feel vindicted

[lightversusdark]

Vindicated.

Re:Microsoft must feel vINDICTED

[Harmonious Botch]

Thanks, but I like the original spelling. It is one letter way from 'indicted'. ( And on slashdot, one letter ain't bad )

Re:Microsoft must feel vindicted

[Master of Transhuman]

That's why vindicted is better - Microsoft is "vindictive" because they were "indicted"...:-)

Wonder if there's a word for "chair throwing"...

We could throw in "vituperative", I suppose.

Re:Microsoft must feel vindicted

[WilliamSChips]

Sellajacent from sella meaning chair and iaceo meaning throw.

Re:Microsoft must feel vindicted

[turbidostato]

"Upgrading is free."

Only if your time values nothing.

"I am running Fedora Core 6 right now and I don't see why you'd want to run an older release"

That only means you don't know better, not that there are no reasons.

Re:Microsoft must feel vindicted

[phoenix.bam!]

Upgrading windows takes the same amount of time as upgrading a linux install.

Do you have any reason whatsoever to run an old version of Fedora or are you just being contradictory? Do you know better?

Re:Microsoft must feel vindicted

[turbidostato]

"Upgrading windows takes the same amount of time as upgrading a linux install."

Yes. And that's (among other things) why so many people are still running Windows 2000 or even Windows NT.

"Do you have any reason whatsoever to run an old version of Fedora"

Not a single one (but currently I don't have any reason to run any version of Fedora, be it old or new either).

But I do have reasons to run old versions of Linux (some other distributions). Main one being "if it runs don't touch it". No matter how easy or cheap is to upgrade it's always more expensive and difficult than to stay the same. But there are others too: legacy software that won't run on modern versions of the OS; legacy hardware that won't run on modern versions of the OS; multipart solutions that will need to be carefull reingineered if you change just a single brick on it.

Any case, I surely know better.

Re:End of the Linux bubble

[RobGarth]

huh? Novell are copping some flack, but they just posted a profit, and Redhat has never been more profitable. Again, I say huh?

Re:End of the Linux bubble

[SirTalon42]

"I am just waiting for Linus to cancel the 2.8 release."

Linus would have to announce a 2.8 release before he could cancel it.

This is a review?

[vbwilliams]

This is horrible...plain and simple.

Re:This is a review?

[vbwilliams]

Damn...wrong thread.

Re:This is a review?

[Master of Transhuman]

Any time you can say "this is horrible" on /. - you're in the right thread.

If you need longterm support, use CentOS

[Nighttime]

From Internet News

Typically a Fedora Core release comes out every six or seven months. Red Hat's flagship offering, Red Hat Enterprise Linux (RHEL), by contrast, comes out every 18 to 24 months. Under the new lifecycle plan a Fedora Core release would have 13 months of support.

"Anything beyond this really seems to be corner cases that would really be better served by something like CentOS for free, RHEL for rock solid support, or Oracle for crackmonkies," Keating wrote. "What does this mean for the "Legacy" project? We feel that the resources currently and in the past that have contributed to the Legacy project could be better used within the Fedora project space."

Re:Justification?

[Anonymous Coward]

No kidding.

Slashdot was the first place I started seeing the mantra "use older distros for older machines" to combat Linux code bloat. So I put it to the test.

An old system (450 MHz K6-3+ saddled with VESA fbdev) under RedHat 6.2, all of its security updates, and a few hand-compiled programs (older ffmpeg, xine-lib with gcc 2.95) can play MPEG4 movies at 512x4xx resolution at about 75% CPU utilization.

The same system, overclocked to 550 MHz and running Gentoo Linux----which I'd expect to perform well with the latest compilers and lean compilation---cannot even play the same file without skipping frames. Evidence that maybe Linux software gets twice slower in 6 years.

Fortunately I have the base install CDs for RedHat 6.2. But the lesson is clear. If old distros disappear they are basically forcing you to throw away old hardware, which is no different from Microsoft. Does it _really_ cost anything more to host old distros and patches, even if they are not actively supported?

Re:Justification?

[LnxAddct]

This has absolutely nothing to do with Red Hat... Fedora Legacy is not controlled or funded or anything by Red Hat. It is community driven, which is what Fedora is all about, and apparently the interest just wasn't there in Fedora Legacy. Hell, at the bottom of http://fedoralegacy.org/ it even says "The Fedora Legacy Project is not a part of Red Hat, Inc."
Regards,
Steve

Re:Justification?

[Martin Blank]

Red Hat has paid programmers, too. But there comes a point when it's time to move to newer versions.

Like a chess move

[theshowmecanuck]

Before we push any pawns or pass blame out, we should try to figure out why the other guy moved his piece where he did. Why is it being shut down?

"The current model for supporting maintenance distributions is being re-examined."

... by the way, we're closing the doors, get out.

That is a bit drastic for just re-examining something. So what is the real story?

I admit that the RHEL theory is a pretty good one. I am curious, are there any stats on how many people *don't* upgrade to new versions of Fedora right away when they come out. If the number is low, then why does this matter, since it really is a cooker type of system for Redhat?

Many don't like it, but Redhat is in business to pay the rent and put food on the table (and a BMW in the driveway and a swimming pool in the backyard if they are able). They are not in it for altruistic purposes. If you can't resolve yourself to this you are being naive and another distro is probably better for you. I used to pay every now and again for a boxed distro from either Redhat, Mandrake, or Suse (kind of like a donation for the work they do), but their customer support is so bad that I decided to just go for the free stuff since I usually fixed problems myself anyway (with free advice fro the 'net). Personally, if I could get good service, I would probably pay for a distro still... But not the inflated prices Redhat charge for what they essentially get for free.

I have been thinking of switching to Kubuntu anyway, so depending on what we find out this might just do it. (I like KDE better than Redhat's favourite Gnome anyway :-)

Re:Like a chess move

[turbidostato]

"Many don't like it, but Redhat is in business to pay the rent and put food on the table [...] f you can't resolve yourself to this you are being naive"

Obviously yes.

The problem is that the people leading The Fedora Project will try to convince you otherwise, that Fedora aims to be an "all purpouse usable" distribution instead of the "red hat cooker" 100% entangled to whatever Red Hat see fits. I find their position quite deceptive.

Different strokes...

[Junta]

There are community projects that do explicitly do Long Term Support, Ubuntu 6.06, CentOS are two glaring examples of making a point of it, and Debian in practice has been that sort of distro. Gentoo, Fedora, et. al are catering to a different, more aggressive sort of mission that somewhat conflicts with the notion of legacy support (frequent releases produce too many overlaps in a long term model, having to maintain 4-5 different trees is not feasible. Ubuntu is doing a decent compromise (6.6 is long term, and they plan to do that once in a while, but still have frequent, shortly maintained releases to acheive the better of both worlds).

Also new hardware

[bill_mcgonigle]

Fedora's whole mission is pretty much implement the cutting edge and let people experiment and play with it. The target audience was inteded to be those desktop users/enthusiasts who would jump on the current release anyway, since it is free. If any business saw a distribution like Fedora and thought it a good idea to base their infrastructure on a Fedora Core release, they are now getting basically what they asked for.

That's pretty harsh considering that they 'knew' Fedora Legacy was out-there.

Most of my clients running Fedora Core releases are doing it because either: a) they needed a newer version of some app than was in a CentOS Release, or b) they have new hardware, the chipsets of which are often only supported in recent kernels which pretty much rules out CentOS. I have one right now with disk controller problems under 2.6.18 with a dualie athlon server we got from NewEgg in August (and wasn't new then, either) that has an nVidia SATA controller - there's a new module in 2.6.19 which addresses it, which means FC6. There' no point in buying a RedHat support license if you're going to compile your own kernel since they won't talk to you on the phone about that machine if you do. I've tried.

Re:Also new hardware

[Master of Transhuman]

Here's the bottom line on your complaint: use the hardware supported by your OS or go with an OS you don't need paid support for.

It's that simple. And little different from Microsoft, IBM, Sun or anybody else in the hardware and/or OS business.

Buying stuff less than six months old and expecting an older server release to support it is not reasonable. Running Fedora Core instead of a server OS to get around this means you get to run new hardware - with no support.

Make up your mind what you want to do. The solution is learn to support your own server OS without paid support. There are plenty of companies doing that. If you're a Linux consultant, this is money in the bank for you.

This is an IT management issue, not a Linux issue (except to the degree that the old problem of lack of hardware vendor supported drivers is raised - which is an issue you have to deal with as a Linux sys admin in every case.)

Re:Also new hardware

[bill_mcgonigle]

Running Fedora Core instead of a server OS to get around this means you get to run new hardware - with no support....Make up your mind what you want to do.

I'm not sure if you're responding to my comment - the grandparent post posited that Fedora was only interesting for Desktop use - I pointed out that people run it for new server hardware. Obviously it's without corporate support - my comment about Redhat illustrates that.

Re:Also new hardware

[Master of Transhuman]

I responded to your post. You seemed to be complaining that your clients couldn't run Red Hat server products because they didn't support newer hardware or newer versions of apps that your client obtained. GP post pointed out that if you run a non-server OS, you can't expect the same level of support - which is true.

My point was that that's how it is with server (and enterprise) software: you either go with what hardware is supported or you try to get around that by running a non-server OS with no support except yourself, the community and Google. Those are your choices, industry-wide.

Oracle won't support its database on anything but supported servers, either. That's the way they run their business. But you CAN run their database on various other versions of the supported distros. You just won't get support from Oracle. Come up with a problem and you'll have to rely on Google, Oracle forums or Oracle consultants. That's not necessarily bad, it's just the way it has to be done.

The issue is how much risk does management want to assume - or pay to avoid. If you can't afford to pay for risk avoidance, use a non-server OS like Fedora and deal with the risk of no support. If you can, don't - buy Red Hat and live within the limitations of the server OS.

The problem, of course, as always, is that client management want to have their cake and eat it, too. That's why they prefer Microsoft - they get to pay for the privilege and take it in the butt via reliability and security and vendor lock-in issues - which are more abstract than supported hardware and apps and harder for idiot management to understand and thus ignored as cost and risk factors. Then, when their proprietary app vendor goes belly up or is acquired by another company and the app discontinued or screwed up, they find themselves in a bind. They forget that this is WHY companies used to develop in-house or via contracted developers - because there weren't any app vendors in their industry years ago. And if their niche is small enough, there may not be any again.

Which is why I recommend companies PLAN for time-developed contracted or in-house development of core applications rather than rely on external vendors. Everything a company depends on for its survival should be second-sourced. Would any company rely on only one shipper? One janitorial service? Companies take second bids for all their critical needs all the time - escept in software.

They assume that just because there is (maybe) more than one product in their critical app market that they can ignore the OS they're running on - just because Microsoft is so big NOW that it doesn't look like it's going out of business. ALL companies are likely to go out of business sooner or later (with the possible exception of Lloyds of London). The only question is will YOUR company go out of business BEFORE or AFTER your supplier does? One advantage Linux has is that it's a movement rather than a company. It's much more likely to survive a nuclear strike on Redmond, Washington, from North Korea, than Microsoft is...:-)

It's braindead, the way companies are managed.

Re:Also new hardware

[bill_mcgonigle]

. GP post pointed out that if you run a non-server OS, you can't expect the same level of support - which is true.

Right, we're in agreement (I think) - I was pointing out that Redhat doesn't offer the same level of support on new hardware which is why it wasn't a valid choice for that application, therefore why some people run Fedora on servers.

CentOS

[KidSock]

Unlike most people I don't reinstall the OS on my laptop every couple of months. Thanks to Fedora Legacy I'm happily running FC 3 and had no plans to reinstall. But now I think I'll have to look seriously at CentOS (RHEL repackaged without the copyrighted material like logos and such). RHEL (and thus CentOS) is supposed to be less "cutting edge" and more about stability over the long term. And because CentOS is just RHEL you know it's going to have more vitality than a community driven project.

Re:CentOS

[Pharmboy]

Actually, since RedHat cut support for RH9, even for those of us that JUST PAID FOR IT a few years ago, I think I finally have the final reason to FINALLY quit RH all together.

I tried to buy the box sets of RedHat, they took my money then changed their way of doing business. I tried to pay them for support, they took the money and ran away without giving the support. I installed FC knowing I would have legacy support, and now I don't. I tried to pay them about $350 a year just to download updates for a few simple servers, and they didn't want my money.

RedHat's policy is simple: "Hey, If you can't afford several hundred dollars per year per machine, then fuck you." No thanks, even MS looks good compared to the AssHats at RedHat. Debian, here we come...

Re:CentOS

[DaveAtFraud]

Planning on doing the same. I'm currently running FC6 on my laptop and primary workstation but CentOS-4.4 on my server. I needed more current hardware support especially on my laptop but I'm thinking CentOS-5 should provide everything I need plus stability, automatic updates to point releases, etc.

I had originally updated from FC4 to FC6 to get the bcm43xx driver for the laptop. The only problem is the bcm43xx driver doesn't work with my AP unless I run the AP fully open (WEP authentication times out before the AP responds then bcm43xx bitches that it got an authentication response when one wasn't expected when the AP responds. S-I-G-H). So I'm still stuck running ndiswrapper for my wireless NIC. So much for more current hardware support.

Finally, CentOS has been stable as a rock on my server while I keep getting random X crashes on both the laptop and more frequently on the workstation (seems to be related to running xscreensaver on the workstation). I don't need that kind of grief.

Cheers,
Dave

Re:CentOS

[Master of Transhuman]

"those of us that JUST PAID FOR IT - a few years ago"

What's wrong with this statement?

I "just" paid for my cell phone four years ago - and today Cingular (who, by the way, bought AT&T FreeToGo cell service a couple years ago, so talk about changing the terms of business!) is refusing to allow any customer to put any more money on the FreeToGo account, forcing an upgrade to their own GoPhone service. The actual cell phone TECHNOLOGY used for FreeToGo is being TURNED OFF on March 31, 2007.

So I go down to the store (at the least minute admittedly, which is how I usually function) and lo and behold, they don't have any of the phones you're supposed to upgrade TO!

So I said "fuck you" to Cingular and switched to a T-Mobile contract with a new, smaller, lighter phone with more features (that I probably won't ever use, but hey.)

But I'm NOT running around saying I "just paid for" my phone...

Re:CentOS

[KidSock]

Debian, here we come...

You mean Ubuntu? If you want long term stable support I think moving to Debian is exactly the opposite of what you want. Debian has way too much religion now and they're behind on updates. I'm still running "stable" on one of my servers and I don't get some security updates for months. I'm itching to move that to CentOS too. I suspect the Debian devs all run "unstable" and basically don't care about everyone downstream. I know there are a lot of Debian fans on ./ but you have to admit Debian just isn't what it used to be. Sounds like the Debian crowd is slowly but surely moving to Ubuntu.

Re:CentOS

[mcrbids]

Thanks to Fedora Legacy I'm happily running FC 3 and had no plans to reinstall. But now I think I'll have to look seriously at CentOS (RHEL repackaged without the copyrighted material like logos and such).

Why?

Even after 6 years of using Linux, it still amazes me how little impact an O/S upgrade typically is. Unlike Windows, where it takes a few weeks to get everything downloaded and installed, and where you always have to put up with a significant amount of data loss, Linux O/S upgrades are relatively painless. Here's my rough procedure:

1) I typically have my HDD partitioned into /home, /boot, and / partitions. So I copy over /etc, /root, and maybe /var/spool over to a subdir on /home, such as /home/xfer/etc...

2) Install the new O/S on top of the / directory partition. Blow away /var, and leave /home alone.

3) copy over any settings from the /home/xfer directory that I find useful. EG: ntp settings, cron jobs, etc.

4) copy over the /root directory stuff.

And that's pretty much it! I can do it all in an afternoon, and have all my settings, configs, and stuff ready to go. I've been using the same /home file system and directories like this since RedHat 6.2. Currently, I'm running FC 6, before that FC 3, FC1, RH 9, RH 7.2, and RH 6.2.

Settings from KMail, Mozilla bookmarks, etc. have survived every single upgrade with very little upset.

Additionally, when I need to swap a hard disk around, it usually takes me less than an hour. For example, I had to ship in my laptop a while back. I popped the drive out and installed it into a desktop machine, It wasn't even 15 minutes before I was up and running with full video, sound, network, etc!

It's true that "weird" hardware common on laptops can cause some troubles with getting everything working on Linux, and Wifi can still be spotty, but it's also amazing how much hassle and fuss I DON'T have to go through when things go wrong!

Oh, well, it's 6:00, wife is calling, it's time to enjoy New Year's Eve!

Re:CentOS

[frdmfghtr]

"those of us that JUST PAID FOR IT - a few years ago"

What's wrong with this statement?

I "just" paid for my cell phone four years ago - ...

I think you misunderstood the statement. The parent post was saying that, a few years ago, he bought a boxed copy of RH9, only to find RH9 support/development cut with little warning shortly after he paid for it. The complaint was not about support being cut four years after buying RH9.

Re:CentOS

[frdmfghtr]

Unlike Windows, where it takes a few weeks to get everything downloaded and installed, and where you always have to put up with a significant amount of data loss, Linux O/S upgrades are relatively painless.

A few WEEKS?? I've gone from booting from the Windows installation CD to back in full operation in a day, including patches, application installation, and data restoration. It's a long day, full of waiting while stuff runs, but it's still a day. Heck, I've started at 9 or 10 at night, and was done by 5 or 6 AM (why in the name of everything that's decent I stayed up all night to do it is beyond me...maybe I'm a closet obsessive-compulsive case.)

Re:CentOS

[rohan972]

I have a similar method for my workstation, but one that gives me more time to get things right. Partitions /home /boot / and a spare partition. I use the spare partition to try out different distos. When I want to upgrade, I install to the spare parition. When I have it how I want it, I copy the necessary files to /boot , mount /home on the new system instead of the old and change the default grub entry. Easy, safe.

Re:CentOS

[DA-MAN]

I think you misunderstood the statement. The parent post was saying that, a few years ago, he bought a boxed copy of RH9, only to find RH9 support/development cut with little warning shortly after he paid for it. The complaint was not about support being cut four years after buying RH9.

IIRC, RH gave users who bought 8/9 a year of support for RHEL WS 3.0.

If in that year, you liked it you could continue to pay for it. Otherwise convert to CentOS or switch to a different distro. RH didn't leave anyone hanging.

Re:CentOS

[mcrbids]

A few WEEKS?? I've gone from booting from the Windows installation CD to back in full operation in a day, including patches, application installation, and data restoration.

IMHO, it usually takes a few weeks to get all the kinks worked out. But let me ask you this:

1) Are you migrating Email accounts from a dozen (or more) users on that system?

2) Data includes the likes of quickbooks? (Many Windows applications do NOT keep their data in your user home folder! This is much less common for *nix apps)

3) Are things like desktop backgrounds, E-Mail filters and the like preserved through O/S upgrades?

I didn't think so...

Re:CentOS

[TrekCycling]

This is precisely why I switched to CentOS a couple years ago and I haven't looked back since. It's absurdly stable (for me at least). It just works and I haven't touched my machine to do any system updates other than "yum update" for the last couple years. That's been very nice. I donated to CentOS and will continue to support them.

Re:CentOS

[Pharmboy]

Although it isn't a popular statement, I don't trust Ubuntu. At any time they can do exactly what RH has done, and I fully expect they will sell out as soon as it will create the greatest financial return.

I was preparing for a migration to SuSE and had already installed it on my personal boxen, and then Novell starting signing deals with MS that are questionable at best.

The problem is, there are many choices if you want to run linux, but there isn't any distro company out there I feel comfortable with, and I am tired of changing distros every 6 months.

As much as MS pisses me off, I will be able to get about 8 or 9 years out of the same platform, and frankly, I am more concerned about running apps, not the underlying OS. I don't think they are shaking in their boots quite yet with all the cluster fuck that is the Linux OS distro scene.

Fedore Core I had better bluetooth support

[lcreech]

That's a shame. Any Fedora Distro after FC1 I have not been able to sync Evolution's address book with via bluetooth with my bluetooth phone. The bluetooth support changed in later versions of Fedora that seriously broke the bluetooth plugins for multisync. It's been over 3 years now and it still doesn't work as well as it used to.

Re:Fedore Core I had better bluetooth support

[RotateLeftByte]

Did you submit a Bugzilla?
Please don't whine here unless you have take all available methods to get the problem rectified.

Re:Fedore Core I had better bluetooth support

[gilboad]

I can't really comment about previous releases, but using FC6/x86_64, I have no problem syncing my Palm T3 * with Evolution and syncing my Nokia 6230 with (x)gnokii.

I also use a bluetooth connection as a poor-man's-wireless connection between my Palm/Ipaq and the Internet.

- Gilboa
* USB sync Evo/Palm is still unstable - hopefully FC7 will ship with a mature pilot-link 0.12. (FC4/5 shipped with beta releases of pilot-link 0.12 and they were severely broken.)

Re:Fedore Core I had better bluetooth support

[lcreech]

It's not a bug, it just isn't supported with the bluetooth support packages that were changed for whatever reason.

Re:Fedore Core I had better bluetooth support

[lcreech]

I have FC6 on several machines. What tools are you using to sync Evolution?

Re:Fedore Core I had better bluetooth support

[gilboad]

Evolution -> Edit -> Synchronization options.
(Which in turn uses gnome-pilot [gpilotd])

- Gilboa

Legacy users ought to...

[drwho]

Legacy users ought to fork the Fedora project in order to support their releases. Surely, there has to be enough talent out there to do this...oh wait, let me guess, you want other people to do the hard work for you, for free? Hrm...better try Ubuntu.

No, seriously, I don't use Fedora so I don't really care that much. But those feeling burned by this ought to unite and take over the legacy support.

Re:Legacy users ought to...

[WindShadow]

The people who want to have this support should either do it, or pay for it, or forget about it. I'm sorry to see this service go away, it means I will eventually have to upgrade, redeploy, or scrap some systems, but I don't have the time to do a proper support myself, and can't justify paying for it instead of a more stable supported release like RHEL.

I'm grateful for the free support I have received, and I might set up a virtual machine for FC3 and rebuild critical applications from source as long as they will build, but I'm not going to complain at losing my free lunch.

Re:Linux isn't Enterprise

[Doppleganger]

As you point out several times in your post, Fedora is not aimed at enterprise usage. The maintenance work at fedoralegacy wasn't aimed towards the enterprise.

What you seem to be missing is that you're looking at a volunteer supported, non-enterprise effort that is closing down, and somehow comparing that to distributions that are aimed at the enterprise and have enterprise funding to support legacy updates. You also seem to somehow be comparing a distribution that issues a new version every few months with bleeding-edge additions with other distributions that are kept stable (and, those distributions are kept stable specifically because they are aimed at the enterprise).

You're not comparing apples to oranges "in some sense". You're comparing apples to oranges, period.

Just one more thing to nudge me back to Solaris

[CapeBretonBarbarian]

This is interesting. I've been slowly moving from the Unix world (Sun/Solaris) to Linux for a while. As part of that, we have been porting applications to Fedora/Red Hat. Lately though, I've been more and more impressed with Solaris 10 (and OpenSolaris). Frankly, I don't see much of an incentive anymore to run Linux on a production server. The only thing Linux seems to deliver better at the moment is x86 driver support and desktop apps. While I don't think we'll necessarily stop our efforts to create a more platform neutral set of applications, I suspect we'll be staying on Solaris for some time. Incidentally, I have no trouble receiving patch support for any of Solaris 10, 9 or 8 production servers. I like the longer support time lines that Sun offers (and much of it for free by the way).

Re:Just one more thing to nudge me back to Solaris

[steve_vmwx]

Just like to point out the limitations of Solaris.

I can see your point in the server space. Solaris does a lot of things very well. As you allude to however, Linux is stronger in the x86 driver arena.

If you're running Solaris friendly hardware then you're fine. If you have hardware that Solaris doesn't like (and I've found it *very* fussy... even with top shelf kit) then IMHO robust Linux distros are the way to go.

For some, having Solaris support dictate their hardware might be OK for the special joy the OS brings. If you're happy to limit your apps as a consequence then obviously your call too.

S

Re:Just one more thing to nudge me back to Solaris

[a.d.trick]

I've never used Solaris before so I can't compare it to Linux; however, IMHO, there are a lot of options available in the GNU/Linux world an my personal experience is that Fedora tends to be one of the worse. There are many other free distros that are much more into delivering a real product that their users can be productive with instead of the treating their users like lab rats.

If OpenSolaris works, than more power to you. Otherwise I'd suggest you try one of these out (In no particular order): Gentoo, Ubuntu, Slackware. Each of them different paradigms, so I can't make a meaningful comparison between them. If you really need Red Hat compatibility, there's always CentOS, but I really don't like a number of ways Red Hat does things.

P.S. I'm not trying to flame Red Hat. They're an awesome company and have done some awesome things with Linux. I just don't like their products.

Re:Just one more thing to nudge me back to Solaris

[that this is not und]

If you're happy to limit your apps as a consequence then obviously your call too.

The NetBSD Packages Collection (Pkgsrc) has been ported to Solaris, along with a number of other non-NetBSD environments. If you can't find what you need in pkgsrc you've got pretty exotic software needs. Read: most, if not all, of the key 'Open Source' packages can be found in the pkgsrc tree.

Obviously, this doesn't all roll out automatically off the .ISO image from Sun. If you can't do a little work in setting up your system, you can always just run windoze.

Re:Just one more thing to nudge me back to Solaris

[Builder]

I'd be on Solaris 10 tomorrow morning if there was a way of running Windows as a guest on there (VMware, Qemu, etc.).
I recently tried to move from Fedora to Ubuntu, and detailed the fun involved here. It would have been easier to move to Solaris if not for the virtualisation thing!

Re:Just one more thing to nudge me back to Solaris

[Blackknight]

I believe Solaris has its own virtualization method but it would be nice if it could run Xen. Xen is one thing that linux does better than anything else, NetBSD is working on it but they don't have LVM support.

Re:Just one more thing to nudge me back to Solaris

[srvivn21]

Just out of curiosity, why are you running Fedora on your servers?

Use RHEL instead. If you don't have the budget, a clone (like CentOS or Whitebox) will perform just as well, but won't have the paid-for support. The version stability of code is there, and you won't have to re-learn how to manage the box. The version of Exim included with RHEL 4 is 4.43. Out of date, but likely still supported... *shrug* At the very least, it comes complete with an exim.conf.

If you want code stability AND fresh software there are Contrib repositories for use with CentOS (see http://lists.centos.org/pipermail/centos/2005-May/ 005618.html for more info), which may very well have the software you are looking for. If not, you are unlikely to break anything by downloading the source and compiling (I took this route with OpenLDAP, as the RH supplied packages were woefully out of date).

For what it's worth, I feel your pain. Under my care are two RedHat 9, one Gentoo, one (seriously out of date) Debian and a whole host of CentOS 4 boxes. Admining the Gentoo box is painfully different. Thankfully, the Debian one is pretty much hands off.

Re:Justification?

[Conor Turton]

Are you aware that Red Hat got lots of code from the people who worked on Fedora for free? It worked both ways.

Re:Justification?

[Anonymous Coward]

My home gateway is an old Pentium 133, with a modest 96Mb of RAM. It runs Debian Sarge fairly well, not a bit slower than the distro that it used to run, Red Hat 9. These two distros have a few years between them, so I must say that Linux doesn't get slower over time, stuff like GNOME and KDE might though...

Re:Justification?

[Anonymous Coward]

333 mghz 333 milligramhectozepto? You're not making any sense.

Re:Justification?

[heson]

Fedora == RHEL Beta

Re:Justification?

[Master of Transhuman]

I'm surprised you can get video playing at all on such a machine.

I have an old Compaq Deskpro 4000 as a second backup machine, CPU upgraded to 400MHz, with 256MB of RAM, and I wouldn't even bother trying to run video on it.

However, I recently managed to get Slackware 10 loaded on it, and while it's hardly spry, it's functional.

This machine used to run Red Hat 7.0 and Windows 98 (still runs Windows 98 - which will crash every few days even if NOTHING is being done on the machine except the wallpaper changer!). I couldn't install any other distro on it because I thought there was an issue with the Compaq BIOS not reporting the disk geometry correctly. However, I discovered the real issue was the later distros were trying to force DMA on the drives, which simply didn't work. By adding ide=nodma to the boot command, I was able to get Slackware to install and run. Seems to run fine, although again, not spry enough for continuous use as a primary machine. 15-second startup for things like Kword and Kspread.

Yes, Linux distros are getting bloated. Compared to Kubuntu, Slackware 10 has everything but the kitchen sink thrown in in terms of preloaded apps.

I'd say it's no big deal if FC 2 is no longer supported. Most machines that ran it okay will probably run a later version or some other recent distro which is supported. Your example of running on an old 450MHz K6 is a bit extreme. Such a machine would more likely have run a Red Hat way before Fedora.

Also, old hardware can be converted to other purposes such as a Linux firewall.

There does come a time when old hardware should be pitched. When I upgrade my current desktop to a newer machine, the current desktop will become a file server and my backup machine - and the Compaq will be stripped for parts and pitched. And good riddance.

Re:Justification?

[pembo13]

Hahah,,,what a stupid comment....as if Redhat has its own magic projects for which it is taking the code of contributors. Dumbass.

Re:Justification?

[Anonymous Coward]

Old RedHat's didn't disappear. They moved them off of the main FTP server and added a README file to give you the new address. Everything back to the 1.0 release is at ftp://archive.download.redhat.com/pub/redhat/linux /.

Re:Doesn't bother me!

[iknowcss]

I started out on RedHat 9.0 back in the day and installed all the newer Fedora Cores as they came out. I thought "man, this is easy. I'm Fedora 4 life!" but then it started to crap out on me, and I couldn't install it on some older boxes, so I tried out plain ole Debian. I did the net install version (a small CD iso that downloads only the packages you request from teh interwebz) and I'm still running it now. It's great. I turned a free donated computer into the (semi)-powerful web server I use for all my web development needs. Try it out. It'll only take a few minutes to download the net install version.

Re:Linux isn't Enterprise

[Master of Transhuman]

Just to point something out.

Just because Fedora Legacy isn't running a patch service any more doesn't mean that those older FC distros are now useless.

The original source of many of those patches are the developers of the software supported by Legacy. Those developers - or other interested third parties - are likely still to develop patches. Those patches can probably be obtained and applied by sys admins who need them.

It just isn't going to be as easy as going to one place and downloading a set of patches, or getting them pushed to your system automatically.

Even on Windows, there are people running projects now that collect Windows patches, put them on a CD or in a bundle and provide a tool to automatically apply them, to make things easier for sys admins who don't want to or can't for some reason use Windows Update Service. I would expect this sort of thing to be done for "orphan" Linux distros to some degree, if it isn't already.

Obviously, as the Legacy project shows, depending a corporate infrastructure on such a service is not wise. But the Legacy situation doesn't mean every distro older than FC5 or FC6 is useless.

Also, even if CentOS goes belly up and there is NO source of RHEL other than Red Hat, well, that's business in the corporate world.

Try getting Windows (or Apple) for free. That's your other choice.

Finally, all this says NOTHING about "Linux for the enterprise". "Enterprises" expect to PAY for their software and their support - not get it free. One advantage Linux has is that you CAN get it for free if you want to and can handle a free OS. But that's not Linux's only advantage. And the other advantages are equally or more important than the simple cost of the OS in monetary terms - even if most CIOs can't comprehend those benefits.

One of the obvious points that you overlooked is that there are enough "second sources" for Linux that an "upgrade" (if not a "migration") is rarely forced. That is not the case in Windows. With Windows, you do what Microsoft says - that's it. That is not the case with Red Hat, SUSE, or anybody else.

The only thing we have here with the Legacy issue is some whining from people who didn't understand the distro they were getting or were using it in inappropriate circumstances.

The proper response: deal with it.

Re:Justification?

[Ngarrang]

Obviously you don't remember the pissing and moaning here about the end of support for Windows 98 and 2000. Lots of /.ers expect everything to be supported forever for free. I still lament the day that support for DOS 6.22 was discontinued.

This is becoming crazy

[hdparm]

Every time the story with Red Hat logo is posted, number of clueless comments increases exponentially. Listen up:

Red Hat is listed corporation. As such, they have to make money to their shareholders and they seem to be doing well, so far.

Huge amount of its workforce time, expertise, money and infrastructure is contribution that RH provides to Fedora Project, free OS of high quality. Everybody is free to join and contribute in many different ways, regardless of technical ability. Although decision making process within the Project was in RH hands, this is changing drastically and Fedora is close to becoming true community effort.

Red Hat made great deal of contribution to wider FLOSS community over time by releasing code, hosting projects, open-sourcing acquired proprietary code, etc.

Fedora Core IS NOT RHEL beta. It makes sense for RH to base its enterprise product on the code tested by wide user base, familiar with RH way of doing Linux.

Fedora Legacy was never RH project. Sure, RH people work on it but that is on their own time. Interest for it vanished. It does not make sense anymore. End of story.

Red Hat is not out there to screw anybody - not you, either. That's what Microsoft and their puppets are for.

If you don't believe this, do join one of the Fedora / RH mailing lists and you'll quickly find out that Red Hat employees are the harshest critics of their own work. Plenty of smart people on those lists, you may even learn something.

Re:This is becoming crazy

[that this is not und]

Red Hat is not out there to screw anybody - not you, either. That's what Microsoft and their puppets are for.

The distinction becomes more and more blurred.

Also: you weren't there to see the icy look I got in response to my question to a high-level representative when Red Hat was out giving seminars in hotel rooms years ago. I simply had asked her if I could share my Red Hat 5.0 CD set with my brother in law.

They've been the closest thing in Linux to chair-throwin' redmondites nearly from the beginning.

Re:This is becoming crazy

[turbidostato]

"Nowadays they'll say how you aren't really running RHEL if you don't have RHN"

Well, if you go and read their EULA you will see quite a bit more. You will see, for instance, that by having one single RHN licensed system you agree to allow RH audit all your boxes and you agree to have RHN licenses for all your RH-based computers.

Re:RHAT just said fsck you to those who do not pay

[Mike McTernan]

Personally, if I had written code that were adopted as a part of a RedHat distribution (or any other), I'd be delighted that the code were being widely distributed and used.

When I've written and released GPL code in the past, I've never had any expectations as to how it will be used outside of the restrictions placed by the GPL; that's part of OSS. Maybe you don't get that, but then, you are an AC - probably never to see this reply.

Re:Justification?

[hemanman]

Just run Windows 2000 Professional on that machine, it works like a charm and the preinstalled Windows Media Player 6.4 was the best version they ever made, if you provide it with some decent codecs.

I run that on an old Dell Laptop 166MHz Pentium(No MMX) with 96MB of ram, if I only use it for viewing movies, Office 2000 work, and webbrowsing using Internet Explorer 5, it is pretty damn decent.

Linux is turning into the same kind of bloatware like games from Firaxis games!!!

-H

Re:Justification?

[Antique Geekmeister]

Yes, it really does cost something. I've seen kernel developers claiming they can backport 2.4, and later 2.6, kernel drivers to their 2.0 kernels "becuause that's what they wrote their modifications in" Never mind that that their modifications are proprietary crap: they think they can backport all the drivers, features, etc. of the newer software releases, including the glibc, gcc, kernels, modutils, sysutils, elfutils, etc. back to the older Os.

Yes, if you paid them enough money, they could install an afterburner on a mule. But eventually, what's the point? The developer OS's need development support, which means rapidly evolving tools, not wasting their time on legacy software and hardware.

Ummm

[nurb432]

Um we are discussion a distribution here, not linux in general. While i wont debate the validity of the general statement that its not ready ( or is ) , reviewing a single *expiremental* distro and generalizing that linux is not 'enterprise ready' is sort of silly.

Re:Justification?

[laffer1]

Part of the problem is that kernel developers need to use larger data structures to handle 64bit computing, larger file systems, and various other requests from the community. In order to fully support new hardware, more memory needs to be used and thus older systems run slowly. You must also consider the complexity of offering several performance hacks for each type of cpu generation. After awhile, just including a basic C function is much easier for older hardware since it gets harder to test over time.

Graphical applications for X11 tend to get much larger and resource intensive as time passes. Developers think they need to add more eye candy than apple or microsoft to get people to use Linux or BSD. Eye candy is not what most of us want.

I noticed that someone mentioned gentoo. I noticed during a project last april that a vanilla 2.6 kernel seemed faster than an gentoo patched kernel on two systems I was testing. Granted I custom compiled the kernel so arguably it should be faster. It could also mean some people are having different experiences based on how much they customized the kernel or possibly how much work went into kernel patches by each distro.

On topic, this shutdown might be a problem for the BSD projects which use fedora core userlands for their linux emulation. FreeBSD had a SoC project to gain Linux 2.6 compatibility with linuxolator so that they could track newer fedora core versions. It might not be an issue for them post RELENG_6_2_0_RELEASE. Other projects might have more trouble tracking newer versions which in turn decreases security over time. I don't see an immediate solution for MidnightBSD.

RedHat 7.3

[hey]

This bums me out. I have been running server very successfully for years with RedHat 7.3.
I've received many updates from legacy. I don't want to change this machine. In fact last time I tried an update Anaconda failed :-(

Re:RedHat 7.3

[Master of Transhuman]

"I don't want to change this machine. In fact last time I tried an update Anaconda failed :-("

What's wrong with this statement?

Time to upgrade, homes.

Re:RedHat 7.3

[that this is not und]

By 'upgrade' it's easy to assume you mean installing something better than a Red Hat product.

I started extensively using Linux back with Slackware and the 1.2.13 kernel.

I, uh, fell off the Linux wagon shortly after my experiences with Red Hat 4.3. Frankly, the 5.0 'upgrade' sucked donkeys. That was when I first started experimenting with NetBSD. I won't say more about what I use now.

Re:RedHat 7.3

[Blackknight]

You can upgrade from Redhat 7 to CentOS 3 without a problem. CentOS 3 to 4 is a pain though.

Re:Justification?

[Billly Gates]

Or install Free or netbsd. Infact Netbsd is highly recomemnd(though I do not use it). They are alot lighter and do not use the bloated glibc.

Also rh6 uses libc which I liked alot better than glibc because it was alot faster and uses alot less ram. The BSD's have not changed much in terms of driver support and code bloat.

Does Redhat See Fedora as a Mistake?

[doodleboy]

I've often wondered about this. Redhat pissed off many of their biggest fans, myself included, with the move to commercial Redhat and free (beta) Fedora. Like many other linux users, I skipped Fedora entirely in favor of CentOS and Ubuntu, etc.

Redhat has stockholders to answer to and did what they thought was the best thing at the time to maximize shareholder value. They're still making money, which is their primary concern, but according to the page rankings on distrowatch.com (not scientific, I know) Redhat/Fedora is now less popular than either Ubuntu or OpenSUSE.

I wonder if Redhat had a do-over if they'd still commercialize Redhat they way they did.

Re:Does Redhat See Fedora as a Mistake?

[burnin1965]

I wonder if Redhat had a do-over if they'd still commercialize Redhat they way they did.

Have you been watching their revenue growth quarter after quarter? 30% to 50% year over year revenue growth for the past 16+ quarters. While their business decision was frustrating for some of us it was definitely a good "business" decision.

While its not the same thing as the previous RHL distro many of us enjoyed working with, Fedora is an excellent distro with a thriving community and with proper attention can even be used where RHL was used in the past.

Re:Justification?

[darkwhite]

The latest compilers, with the default settings, produce code that's bloated for the tiny machine you describe. I suspect GCC 4 with "-Os" would do better, as long as you were going Gentoo anyway. Also, what happened most likely was that your video setup was suboptimal and had minimal or no acceleration. In general, I find it very hard to believe that a system with the later 2.6 kernels, GCC4, and properly configured video and multimedia packages would have trouble beating an old 2.4 system, especially if it's Gentoo, not saddled with a bunch of services starting by default.

Re:Justification?

[that this is not und]

I can close my eyes and try to visualize what kind of tool would have called the Microsoft 'support line' for issues they were having with DOS 6.22. Nope. I can't think of anything. Now, Windows 3.1 or 3.11 running on top of DOS I can see.

Then again, I used to work with an engineer who seemed to live to sit on hold in tech support calls to all the half-assed companies from that era. I think he even harped on tech people from products like 'PopUp DOS' the free application manager they used to bundle with Logitech mice.

Re:Justification?

[that this is not und]

I can run the latest NetBSD release on my Macintosh SE/30 and get actual servicible use out of the machine.

Mind you, that is with the Tab Window Manager (twm), not the latest croftly kludge WMs rolled together using object-oriented code glop. And about what you can do with an SE/30 with that OS is maybe play some GNU Chess graphically under X, and use it as a terminal to connect to other boxes. But an SE/30 is a really REALLY old machine.

So...

[rodgerd]

I can either run my home Fedora 3 server unsupportable, upgrade to a recent version of Fedora (a fairly fraught process if the crappy FC 1 -> 2 -> 3 process is any indication), or bite the bullet and go to Debian or Ubuntu server.

The answer isn't looking good for the last RH box in my house.

Re:So...

[NerveGas]

Or use CentOS, which keeps your redhat-ish knowledge useful, and have a good lifespan of updates available.

steve

Ubuntu and other Debian derivatives

[leighklotz]

I've been using RedHat since 5.0, and remember the great service they did with the 1.0 disk drives pre-loaded you could order.
But lately it seems like RPM has stagnated, and Fedora appears to be confused as to its direction, and finally, the upgrades have been getting harder lately. Ubuntu is one of the first Debian derivatives that seems to work as well as RedHat has done, with well-done installers, etc. I suspect that with the handwriting on the wall about upgrade support from the Legacy project, my next upgrade will be to Ubuntu.

Re:Ubuntu and other Debian derivatives

[catchy_handle]

When RedHat decided my $60 wasn't good enough and went 'strictly commercial' I went back to SUSE (then SuSE). Gave Fedora a pass out of protest. SUSE went to shit, what's left--Ubunto, which is sorely lacking. Sound broken, hot-plug usb media reader broken.

We're going backwards. You can't buy Linux at Fry's, Circuit City, Best Buy or anywhere, WTF!

I'm glad to hear RHEL is making money, bully on them. It was a sad/fateful day when they pulled the plug on their retail distro. My New Year's wish is for them to reconsider their userbase. Their success was built by hobbiests that grew up to be IT professionals and sold their product for them. Why'd they pull that plug?

Kyle A.

Why Would Microsoft Feel Vindicted?

[EXTomar]

Fedora is supposed to be a bleeding edge distro of Red Hat technology that is frequently rebuilt and released. Fedora Legacy, although well intentioned was sort of misplaced for this technology Why would you want legacy support for a very experimental software distribution. If you need real enterprise stablity you should be using true Red Hat Enterprise or CentOS. I would never use Fedora for any server or workstation that I wouldn't want to manage rigorously.

Fedora does a lot of good for the Open Source Community as well as Red Hat itself but lets not mistake it for a long term solution to enterprise technology. If Fedora Legacy wants to learn anything I think there is a place for a Fedora that has less agressive upgrade path. They should do what Red Hat does with Fedora which is build from the best of Fedora. Maybe they something try for a distro that mimics an "even release" sort of way.

Re:Why so many variations of kernel and apps?

[JoshJ]

"hunting down a binary package" consists of either 'apt-get install vim' as root on Debian-based systems, 'emerge vim' on Gentoo, 'yum install vim' on RPM-based systems (basically everything else except slackware and linux from scratch, and I think you can even set those up to use apt or yum.) In most cases, there's a graphical frontend. The GNOME package manager in Debian-based systems is basically "check box to install program". Checkbox next to firefox, evolution, OO.org, etc, etc.

heh.

[Snipes420]

fork what you want
fsck the rest

happy new year.

Re:Doesn't bother me!

[RotateLeftByte]

I have to take issue with your statement about the default partitioning layout.
I have installed FC6 on at least 6 different physical systems including a Mac Mini(ppc), a Dell 8600 & 8100 laptop and an IBM T43P laptop and I NEVER use the default partition layout.

I have never had a failure at all. I use /boot / /usr /var /opt /home & swap partitions.

I will say that all RedHat installers do not calculate the partition sizes correctly. So, if I want to be user that they are set to proper multiples of sectors etc I format the disk(s) using a live distro first. Then in FC6 (or other distro) I select the partion and the njust edit it to tell it the mount point and filesystem type.

As far as the software selection process goes it is IMHO a lot better than SLES 9 where you can get into a situation where because of dependancy failures the installation can't be progressed.
You could always just install everything btw.

The Fedora project does not hide the fact that it is a leading egde distro. It is a test bed for a lot of software that finds its way into RHEL and lots of other distros over time. Many of the Fedora Users actually enjoy being lab rats.

Finally, this whole article is not news. It was covered on OSNEWS several weeks ago(mid december).

Oh no! it is going to affect many blogs

[victwu]

This decision is bad! There are many websites, tech articles, blogs like mine [Ocean Of Technologies] http://victwu.blogspot.com/ that reference to the link http://download.fedoralegacy.org./ I spent quite a fair bit of time on this blog and now I have to keep an eye on when the fedoralegacy goes disappear. Can anyone suggest me the alternative reference? Victor Wu

I dont know what 2 say

[vz3phyre]

It seems like fedora had too rapid new development i think.. within 6 month (if we want to be updated) there new release.. Its good but for newbie like me upgrade FC seems a difficult process bcoz of the hardware problem etc... but their support are great.. i've solved many problems by refering to them.. too sad this is happen and i been using fc4..

So, may be after this i will consider to openbsd but it seems more difficult to configure. My last word here .. i donno what to say :(

Why I still use Fedora

[Jump]

Just yesterday, I upgraded my home PC from FC4 to FC6. This went smoothly as always (I did upgrades since RH5.0).
It took me half day, however, because I first copied the system to a new hard drive in case something goes horribly wrong. I wish anaconda had a way to first clone installations before upgrading. Using fdisk and cpio works, but is not the simplest way for everybody. As usual, many 'rpmnew' files are installed and need to be sorted out manually. For a mission critical system, I would always migrate like this. Making the system dual boot in two identical copies of the system, then upgrading one of them.

So why not CentOS or RHEL (the later we use at work)? I like to tinker with things, really. This way I keep up with the changes in Linux and the knowledge is important for my work. I used debian before, and was quite disappointed by it's old software (woody). Probably a good choice for a cheap server running on old hardware. Used SuSe for a while and didn't like some aspects (i.e. yast2).
What I like about Fedora is, that they try out new innovative ideas (i.e. stateless linux). Yes, occasional you get bitten and you have to upgrade often. I wished they would give up the point release idea, and instead make incremental updates.

Having only 13 month of security updates is not great. It's a pity that legacy is shutting down. This also kind of shifts the balance between stability (patches) and quirks (new features) in the wrong direction. We will see how it goes, or if Fedora starts to decline. At the moment the quality is quite high in my opinion. It also reminds us that nothing is guaranteed: Fedora Legacy originally promised to extend support to older versions but they could not keep up with the large number of releases. Maybe they should focus on one or two old releases.

If I would run a server at work with FC4-, I would now upgrade to RHEL or CentOS (but I would rather prefer to pay for the redhat network service). Running a server with FC makes only sense if you have the time to upgrade it every year -and- you need some cutting edge software / hardware support on it.

Running FC on older hardware? Not really a good idea. I have an older laptop running RH9, because it cannot upgrade (not enough memory to run anaconda). Sometimes the best solution is to uninstall the software packages which creates security problems.

Too short

[djfake]

I've typically skipped every other version of Fedora and only installed the "even" numbers. The point is, it's not practical to reinstall a desktop os every six months. Not even if your a geek with nothing to do. everyone bags on Microsoft for keeping XP around so long, but I for one am in no hurry to upgrade or reinstall an os "for fun". It's a lot of work, and the cost/benefit is rarely ever worth it.
this is something Linux distros in general should think about. there will never be mainstream acceptance if the product life is measured in months not years.

Re:Justification?

[Lucas.Langa]

still runs Windows 98 - which will crash every few days even if NOTHING is being done on the machine except the wallpaper changer!

Maybe it has something to do with the wallpaper changer then.

Re:Justification?

[Master of Transhuman]

Nope - has to do with Windows 98 running out of "resources" - a known issue with 98.

It's the Windows 98 equivalent of "memory leaks".

Re:Justification?

[ajs]

The BSD's have not changed much in terms of driver support and code bloat.

bloat == functionality, in this case. Applications under Linux tend to be smaller because there's a shared library that supports more of the functionality that they want. Under the BSDs, this functionality needs to be replaced on a per-application basis by code that doesn't even have to exit the pre-processor on Linux.

It's a trade-off, and both are valid. I happen to like the way Linux works out, but there's nothing wrong with going the other way if you understand the scaling implications.

Re:Justification?

[Ganniterix]

Bull$h|t!! Back in 1999-2001 I used to run a similar system, played mp3's, some videos and did all the hell I wanted with 128 MB of ram. I upgraded it to 256MB and did all those things on Windows XP.