Marcelo Tosatti released the official 2.4.29 kernel, unchanged from 2.4.29-rc4. Among many other enhancements and bug fixes, this release includes a number of local and remote security fixes, and thus is a recommended upgrade for users of the 2.4 kernel.
Geez, why can’t they simply stick w/ 2.6?
because there are STABLE servers that NEED 2.4 on production
even 2.6 is stable enough there are others in this world that need 2.4 if you don’t great
Apparently this bug still afflicts 2.4.29…
http://isec.pl/vulnerabilities/isec-0021-uselib.txt
Not everything is on 2.6 yet. We have a lot of linux appliance stuff that is still on 2.4 until they are 100% sure they’ll have no issues and finish testing to 2.6. So, these little fixes do come in handy.
>2.4.x SUCKS !!!
>2.6.10 is the best kernel
Just becouse it’s better you can’t drop it. A lot of production servers and distributions still use 2.4 so it’s good people keep updateing it.
As far as I’m concerned, the Linux kernel 2.4 is still the STABLE version (very much like FreeBSD 4.x is STABLE and 5.x is not). The kernel 2.6 introduces many new features (ALSA, HAL, etc.) that simply don’t work with all hardware as expected. I’ve worked for Linux support so I damn well know these things. The 2.6 kernel just isn’t stable yet.
In response to:
> Apparently this bug still afflicts 2.4.29…
> http://isec.pl/vulnerabilities/isec-0021-uselib.txt
Actually, that bug afflicted 2.4.29-pre3 but was fixed in the creation of 2.4.29-rc1. It went to rc4 before being released, so the released version no longer contains that bug.
I have a box where I play enemy territory on (a p4 1.6 gf4ti but only 128 megs of ram)
In 2.4 the maps take about 30 secs to load but they do load and I can play on most servers without issue.
In 2.6, the first is about 30, but then they take longer and longer until literally in the minutes ( a map taking 200 secs to load) and I get kicked from the server.
And ya, I’ve made sure about the nice values for X and all that. And yes ram is cheap, but that’s besides the point.
The point being here is a situation where 2.4 handles it fine and 2.6 chokes. It’s somewhat concerning about how inefficient 2.6 seems when put in a situation of low ram.
The 2.6 kernel series is labeled ‘stable’ by kernel maintainers. As far as I’m concerned, Blowmywhistle is right in saying that just isn’t true. Well, depending on your view of ‘stable’, ofcourse.
I’ve been building my own Linux system from sources for a while now, a la Linux From Scratch, and with updating 2.4 kernels it was usually a matter of going through the release notes, bumping the version number, and recompile.
With 2.6 however, every single ‘stable’ version I’ve used, had at least some issue or another. No showstoppers or anything, but always something. A typo here, something small broken that worked before, etc. Small issues, but then again, that’s just what I personally ran into, and I’m just 1 person, and not even a developer. Says enough about the state of the whole thing, I’d say.
Some people argue that 2.6 series kernels are already more stable than 2.4 for production use, better capable of handling heavy loads, scaling better with huge workloads on many-CPU big iron, etc. I’m not even going to argue against this, it may well be true. So far I *never* experienced a hard lock-up under Linux, and 2.6 kernel series hasn’t changed that (for me). My personal impression of 2.6 is “very capable, powerful on the right hardware, but still needs some more polishing”.
But stable? NO. If you have the right workload and known-good big iron to run it, it may run even better/solid than 2.4 series. In that respect: stable.
But otherwise, I call stable when ordinary users can’t find issues, when changes between final releases don’t break any stuff that worked before, and upgrading is down to bumping version & recompile. And 2.6 series clearly isn’t there yet, no matter how well it is working for you. Maybe that’s why a 2.7 development branch hasn’t been forked yet?
Until Torvalds leaves kernel 2.6 alone and starts kernel 2.7 I will not see kernel 2.6 as anything more than a ploy by Linus to get a bigger testing audience on a dev kernel. Something that Linus has always stated needs to happen for Linux to improve faster. Not a bad thing at all, I for one am for it. But not on my servers . . .
I use kernel 2.6 on workstations and non-critical infrastructure, but I just don’t trust Linus not to rip out a major part of the kernel and replace it ala the VM fiasco with 2.4. Linus is a great developer and provides the perfect lead for development kernels, but he’s stated time and time again that he gets bored with stablilizing kernels. That’s not what he wants to do, he would prefer bleeding edge stuff on the latest dev kernel. This is why he USUALLY hands off the stable kernels to others.
That said, kernel 2.6 is progressing nicely. Definately a piece of work by all involved.
You might try Con Kolivas’ mapped watermark patch, it evicts RAM before full load instead of waiting.
http://members.optusnet.com.au/ckolivas/kernel/
I mean, I use 2.6, I’m more or less stuck with 2.6, but I don’t think it’s stable.
I am scared to upgrade to newer versions of 2.6 for fear of what might happen (from horrific disk corruption with SATA in 2.6.3) to (inexplicable stuff like hangs during CD reads with no logging output – I get these with 2.6.9) – or having stuff like cd recording break (e.g. between 2.6.6 and 2.6.8).
Stuff like the ‘ub’ driver. Not marked as experimental in the kernel config, but plainly unusable – it kernel panicced my machine on device removal every time. How this slipped through is a mystery to me.
Functionality that was in 2.4 is still not in 2.6 (robust firewire support), and while generally I think things are getting better functionality-wise, the 2.6 development approach is leaving me, as a user who likes to compile his own kernel and track the latest updates – pretty unimpressed.
Now I realise the developers need to be free to do things their way, and I dont expect the world to stop because i’m upset about my machine silently locking up all the time, but it is troubling to simply not be able to trust the kernels you download from kernel.org any more.
Right now I have a couple of servers which cannot go to Kernel 2.6 because of driver issues. Hopefully I can resolve it but until then I’m keeping them under Kernel 2.4. Note I always update my 2.4 series kernel which my Red Hat 9 server is running on.
2.4 runs all my programs and hardware great in WINE
2.6 barely detects my hardware (older gear) and runs WINE, what seems to be in super-slow motion.
As far as my desktop Knoppix is concerned, I ONLY boot into 2.4 (2.4.27 to be exact.) Wlan and Sound work only in 2.4 for me, 2.6 breaks both.
2.4 also FEELS faster, for whatever that is worth, on a celeron 1.1ghz with 256mb of ram.
It’ll be a long while before I upgrade.
I agree with the opinion that 2.6 still has some teething issues, but what many of the posters here forget is that many people held out and continued to use the 2.2 series for a long time for the same reasons that many are now holding out and using the 2.4 kernels.
Hell, some people still are on the 2.2 kernels. 2.6 is a much better foundation than 2.2 and 2.4, it just needs about another 9 months to be perfect.
Having said this, I am using it with close to no issues on most of my computers. I think the distributors themselves have to do a better job of stabilizing Linux and having fewer and more meaningful upgrades. We need to move to 2 year cycles, instead of six months.
the kernel 2.6 introduces many new features (ALSA, HAL, etc.) that simply don’t work with all hardware as expected. I’ve worked for Linux support so I damn well know these things.
—-
hal, if you are talking about hal.freedesktop.org is purely a user space abstraction.
good news 2.4 has been updated, upgrading to 2.6 always has given me problems, lan wont work, get lan to work, sound wont work, my nvidia card locks randomly in 2.6 whereas 2.4 doesnt lock,and works with all my new hardware without issues.
Thanks for correction. Still, my main point is valid and the new technology introduced with the 2.6 kernel needs to mature a bit before it can be called stable. I don’t really care whether the actual problems in moving to the 2.6 kernel series are due to problems in the actual kernelspace or problems in the interaction between kernelspace and userspace. The fact remains that many users find moving from the 2.4 kernel to the 2.6 kernel problematic. Things that used to work with the 2.4 suddenly don’t work with the 2.6 kernel. In this sense 2.4 is the stable kernel and 2.6 is far from stable.
Does anyone know if 2.4.29 support SATA drive? How about support for SATA in 2.6.xx?
It seems that many distros have problem with installing to a SATA drive. At least some will not install without doing extra work.
Thanks.
It was added to the 2.4.27 kernel, so it should work. My SATA drive has worked fine with both 2.4 and 2.6 kernels. I did however have a problem with installing FC3, but the problem may be just with something they changed and not affect other distros, it certainly hasn’t affected any kernels I built myself.