Slashdot Mirror
Last 2.5.x Linux Kernel Released
Kourino writes "Today on LKML,
Linus released 2.5.75, which he said will be "the last 2.5.x kernel from me", and that he and Andrew Morton are going to start a 2.6-pre series soon. While this certainly does mean things could get interesting soon, don't hold your breath about seeing the actual 2.6 for a while; there are still many areas that need work. This essentially means that the development branch is going into maintenance mode, and new features probably won't get in after this point. Changes of note in 2.5.75 include a merge of the anticipatory scheduler from Andrew Morton's -mm tree and updates from several architectures."
What about Reiser4?
Will the 2.6 "stable" kernel series actually be stable?
The 2.4 series had this public cloud of wierd problems hanging over it its entire existence. It seems like 2.4 never really seemed "trustworthy", they kept making huge and highly experimental changes and 2.4 seemed just kind of like a work in progress for its entire run. Will 2.6.0 be totally safe to download and run and install in a production environment, or is that going to be kind of a "well thats still sort of experimental be careful"? And if the latter, why the heck aren't they staying in 2.5 until it's ready for production.
Am I just too paranoid, or do you know what i mean?
-- anonymous and terrified
The SNARE folks say they are working to get C2-style auditing capability in the kernel, since the old hooks were broken/fixed in 2.4.21. This is a big feature that is keeping Linux from being a "serious" player in "secure" environments, such as certain government-controlled areas.
>Just name it 2.6 - everyone will flock to it because 2.even means that it must be a stable release, never mind it's the first release.
And those who made that mistake with 2.4.0 will continue to ignore 2.6 until it's proven itself stable and not find the bugs anyway.
(I'm not one of them, but I have time to spend
on following dev releases. Not everybody does).
I'm not a fan of the "it compiles, ship it! and we'll fix it in a service pack" mentality.
- Muggins the Mad
I have a feeling I am going to be modded as a troll but I have karma to burn.
I use to be a big fan of linux but the latest 2.4x came with a bad vm in the so called stable release branch and I heard of dismal uptimes for smp systems with 4 or more processors. Infact Debian still uses the 2.2 kernel by default because of the bugs sorrounding 2.4
I am no longer in IT but if I was I would be more in favor of FreeBSD. I heard 5.0 is alot more scalable then the 4.x branch.
Anyway its reputation for those who are not Linux fanatics on slashdot will be better. Linux 2.0 was rock solid. However the quality has gone down hill recently. Yes Linux 2.4 can scale quite well but in real world uses filesystem corruptions, xinet freezing, and kernel panics happen on smp hardware.
Since Linus now wors at OSDL he can now test these features on high end hardware. Linux is stable on pc class hardware but that is all most kernel hackers have to test the kernel.
http://saveie6.com/
Here's a good pointer: Dave's "post-Halloween" doc.
I host a bunch of VPSs based on Jeff Dike's UML (User Mode Linux) project.
One (of the many) cool things UML allows is for you to try out new kernels without having to dedicate a real box to it. Even if you're only dedicating the box to it between kernel swap reboots. Especially if you're not sure if the new kernel will corrupt your precious partitions.
The UML 'host' server can continue to run whatever stable 2.4 kernel you need (in my case 2.4.21).
You SSH from your 'host' server into your hosted UML kernel. Play around, test reliability, fiddle with new features, regression test your apps.
So anyway, I'm off to grab the new kernel and have a play. Maybe even see if there are any crazies out there who want hosting with the 2.5/2.6-pre kernel.
- Peter
RimuHosting - Linux VPS Hosting
I'm not a fan of the "it compiles, ship it! and we'll fix it in a service pack" mentality.
And I'm not a fan of waiting for beta testing that will never happen before releasing it. It is a thousand times easier to find bugs that have been found. Therefore the method that allows you to find the most bugs in the shortest amount of time is the best method. This is assuming that you are not actually selling the product for a profit. In other words, release the 2.6 kernel because no one important is going to use it until it gets put into a distribution. So there's sort of always a testing period after the release but before most people start using it.
ok, I've managed to completly disagree with myself several times. I guess that means I must be right. Its pretty clear to me that there *isn't* a best solution. So for heaven's sake just do something, it will be better than nothing.
Well.. maybe. Or Maybe not. But Definitely not sort of.
If the Amd world view of how to achieve 32 bit without emu on a 64 bit platform are to fly then the adoption of AMD by the server world is essential for Linux in the future. Blindly following the Intel/MS lead may lead to kaos. The same as blindly imitating Microsofts functions by reverse engineering, is for programmers.
The office desktop lock of MS is not the route that Linux should take, the applied advanced scientific computing and clustering is the best route. When a great scientific workstation can be had for the price of a Linux install on a 64 bit AMD system the business computing world will finally start to wake up and take notice.
OH THE SHAME I fell off the wagon and use sigs again!
NEW XCONFIG!!
x t
Check out: http://www.codemonkey.org.uk/post-halloween-2.5.t
Now, when someone does make xconfig it uses the qt libraries. There is also a make gconfig for all you gnome people. While I like the advancement, it's annoying that even at the deepest level, the kenel, people are forced to repeat functuality for different libraries. While I love the choice, it is just annoying that we so much redundency for these libraries. It seems that programmers are programming more for the libraries than they are for the users. Unfortunitly, I can't think of a way to solve this.
However, it does suck for anyone who uses another window manager and doesn't have/want qt or gnome. I guess they have to live with ncurses.
Only dead fish swim with the stream...
I use RH, but I stop using it at the ide-SCSI level. I use a locally compiled MPlayer CDRtools at the latest release level and XCDroast. Its a bit more fiddly to configure, but it works well on my older machines. I also have it on a true SCSI machine and again it works ok.
See my journal, I write things there
As for scaling past 4 CPU's, that was much more true before 2.4. While it's still not going to scale linearly to an infinite number of processers, improvements along the way in the 2.4 series and the o(1) scheduler have helped out quite a bit.
I do believe that much of the non-scaling of Linux past 4 CPU's is, to some extent, actually the fault of the hardware, as the great majority of multiprocesser hardware has bottlenecks that impeded linear scaling.
As a simple example, take a look at the dual P4 Xeons on the 533 MHz bus. Sounds spiffy, right? Well, you're splitting the 533 FSB and memory busses between both processers, giving each one an effective *266 MHz* when under load. Seeing as how even the slowest single-proc P4 has a 400 MHz FSB, you can see off of the bat that you're hitting bottlenecks!
Look at the AthlonMP series: Each processer has it's own independent bus! However, the only available motherboards have a single-channel, 266 MHz memory architecture. That gives each processer only 133 MHz effective memory bus under load. The simple addition of a dual-channel memory controller (like the one on the NForce boards) would give the AthlonMP's a real shot in the arm.
Now, it may seem like I'm just talking about low-end multiprocesser machines, but here's another example: Even on some of the higher-end machines, there are restrictive bottlenecks. By naming names, I'll only start a flame-war with the zeolots, but suffice it to say that there are $35,000 "high-end" servers that have *less* total memory bandwidth than that $3,000 dual P4 Xeon. That is pretty pathetic!
It's pretty easy to see why someone who didn't realize that could plop $35,000 on a 4-way, big-name machine that had less memory bandwidth than a $3,000 dual P4 machine, see that under load both of them performed similarly, and say "Well, Linux must not scale well."
To make matters worse, the kind of applications that are run on multi-CPU machines tend to be things like RDBMS', which do not lend themselves well to clustering. Here's the catch: Those types of applications tend to be the most memory-demanding. So, take a single P4 with a 533 MHz bus, and install your RDBMS on it. Take a dual P4 Xeon with a 533 MHz bus, and try the RDBMS. You're certainly not going to scale linearly, but that's because you still haven't improved the memory bandwidth.
steve
Oh, you're not stuck, you're just unable to let go of the onion rings.