2.6.38 Kernel, will it be included in Tumbleweed? Will it make a differerence for the average user?

Saw this article Linux 2.6.38 Boosts Performance - InternetNews.com and the title sort of jump out at you "Linux 2.6.38 Boosts Performance New Linux kernel takes big steps forward to make the open source operating system faster than ever"

my 2 q’s:
will it be included in Tumbleweed? The description of Tumbleweed would imply yes but in the past openSuse never up’ed the Kernel to a new major release once a version was shipped, but then Tumbleweed is playing under new rules.

Will it make a differerence for the average user? as I use VirtualBox it might?

regards,

I recommend the average user stay away from Tumbleweed.

Or only if the average user has a spare PC, then play with Tumbleweed.

Maybe after a couple of years have gone by, and we have a better feel as to the success of Tumbleweed, will it IMHO be a good place for the average user. Just IMHO.

From my limited experience of using Tumbleweed repo on 11.3 during that testing period, I upgraded kernel-desktop from 2.6.34 to 2.6.36 and 2.6.37, so I see no reason why it won’t reach 2.6.38, starting from 2.6.37 on 11.4 now.

are you still using it? and if not, why not? would agree with oldcpu’s concerns? Considering I have the dvd and could always "repair " and replace the kernel to the default …

There is another alternative to test the 2.6.38 kernel. Update /etc/zypp/zypp.conf to allow for multiple kernel versions:



##
## Packages which can be installed in different versions at the same time.
##
## Packages are selected either by name, or by provides. In the later case
## the string must start with "provides:" immediately followed by the capability.
##
## Example:
##	kernel				- just packages whith name 'kernel'
##	provides:multiversion(kernel)   - all packages providing 'multiversion(kernel)'
##					  (kenel and kmp packages should do this)
## Valid values:
##	Comma separated list of packages.
##
## Default value:
##	empty
##
# multiversion = provides:multiversion(kernel)
**multiversion = kernel-desktop
**
##

Add the Kernel/HEAD repository:

http://http://download.opensuse.org/repositories/Kernel:/HEAD/openSUSE_11.4/

If you are running 11.3, use the associated 11.3 kernel/HEAD repo.

Add the 2.6.38 kernel via YAST (or zypper) and re-boot. If the new kernel does not work satisfactorily (or fails to meet your expectations), you may simply re-boot, selecting your previous kernel from the GRUB menu. Of course, you may delete this new kernel (2.6.38) if your are no longer interested.

I have used this method with very good results, and without requiring more than a re-boot to return to current state. I suggest that you add NOTHING ELSE from the kernel/HEAD repository until you are sure that you wish to keep the kernel.

Further, my experiences with the 2.6.38 kernel, from -rc1 to -rc8, have varied from exciting to disappointment, and finally resigned acceptance. It has appeared faster, although has not been without quirks. The -rc6 version crashed and burned in early boot, several versions have been snarky on ATI graphics, and (most disappointing of all) the new and improved kernel support (NOT the Xorg support!) for the Intel GMA graphics showed up in -rc1 and -rc2, regressed in -rc3 and remains somewhere by the wayside.

I currently run 2.6.38 on 11.3 and 11.4 installs, both on an Intel i5-430 with Intel GMA graphics. As for the ATI graphics, I still test 2.6.38. I have two problems (which I believe are due to KDE) that may be sorted after a re-install. Once you successfully install 2.6.38, do a test re-boot back to your current kernel. THEN you will notice the difference!

I forgot to add that I used it on 11.3 test partition as Tumbleweed was only testing its repo. I continue to use and update/upgrade that system, and have had no problems over several months of carefully upgrading the kernel through the levels mentioned above. Of course there were changes in Kernel support for my hardware, but no upgrading issues. Other than that, on 11.3 I didn’t see any other updates in Tumbleweed. Therefore once I’m happy that the packman repo changes and package upgrades/downgrades have settled down, I will begin testing Tumbleweed on 11.4 (there is already a smallish kernel upgrade in the repo). Tumbleweed on 11.4 should mean much more package upgrading than just kernel packages! That testing will continue for months before I would even consider a decision about moving to a rolling release for a main system. Standard 11.4 testing is looking very good here so I will migrate my notebook to that within weeks if nothing adverse happens. :slight_smile:

Update /etc/zypp/zypp.conf to allow for multiple kernel versions:

That is worth mentioning for Tumbleweed as well. I did that for my 11.3 Tumbleweed kernel upgrades, and would definitely recommend it for 11.4 tumbleweeding. :wink:

As for the original question:
the 2.6.38 Kernel should NOT make a big difference for the “average user”. It is actually speeding up not the desktop itself but it allows for a speeder desktop in case you are running cpu intensive tasks in a terminal in the same moment, while a desktop application competes for cpu resources. So if you use practically only KDE or Gnome apps then you will not have any big advantage as far as i have understood. However, if you use for examples programs running in shell environment while working on the desktop (compiling kernels, running matlab statistical software, running any other kind of software that is CPU intensive and requires lots of reckoning power), in this case the 2.6.38 kernel should AFAIK make some difference. It is not a booster of CPU or hardware performance. So do not expect to make a lame old duck fly like an eagle. It will just make sure your desktop stays responsive as usual without the well known signs of lack of resources (sluggish performance, slow change of windows, etc).
Tumbleweed: I do not use tumbleweed (as I am still with 11.1) but when the kernel HEAD repository was still available for 11.1 there was the rpm version for 2.6.37. I did install it, worked flawlessly, you may have minor issues. I think one should use a new kernel version that is not the one of the distribution when the hardware one runs (like in my case) is too new or does simply not work without the new kernel installed. If you stay with the official kernel you will enjoy regular security fixes and will (should) have more calm when performing the upgrade as it is supposed that these versions had better testing on the average system.

Interesting, I did not know this about running multiple kernels

question regarding using the head repo (and I assume works for Tubleweed also):

as I compile the Nvidia drivers the “multiversion = kernel-desktop” line in /etc/zypp/zypp.conf would allow installation of all needed packages required to compile Nvidia for the 2nd kernel, the system would be smart enough to select the proper source & header files required for the compilation and grub would allow me to login to which ever I wish? Cool! Learning is good!

ps - I’m not jumping in to do this, just trying to understand what’s involved and the risks

appreciation to all the responders,

Kernel:HEAD repo is bleeding edge like Factory, without “emphasis on stability” that Tumbleweed has. I don’t see why the more regular kernel upgrades in Tumbleweed will not have the appropriate security fixes already applied, and the same for other packages there.

On the standard openSUSE system, most packages never get upgraded between distro releases. There are occasional kernel updates (security and other patches) and the same for some other system packages, but Packman packages seem to be upgraded/updated more frequently. The objective of Tumbleweed is to have more frequent but stable package upgrades. This will come with an increase in risk, testing/troubleshooting, admin, and maintenance overhead for the user responsible.

Decided to give 2.6.38 a go. Installed the version from factory (2.6.38-rc7-2) on a clean install of 11.4.

Well it booted up - always a good start! First impressions is that it does seem faster (admittingly on a reasonably fast p/c) with applications.

Look, that is by no means a thread about the usefulness or stability of Tumbleweed (that in all cases will have occasion to show its validity through time, no?) but a response to the original question: is the kernel 2.6.38 making you spiderman or wonderwoman? So my answer was about this, there was no intent to say: Tumbleweed is no good idea or intrinsically instable. Just to avoid unnecessary itches.
However, if Tumbleweed has identical stability of the original distro kernel and same security fixes…why do we still bother with the original repos? I understand the project and (for me as user) I do approve it. But the thing was about the “average user” (whatever that may mean) and her/his relationship with or advantage to perform a kernel update for the aforementioned advantages.

For the record, as of today 15-March, I think tumbleweed’s “stable” kernel for openSUSE-11.4 is the 2.6.37.3-16 (updated from the 2.6.37.1-12 that comes with 11.4). … IF I am correct in this then it may be a short wait before we see a stable 2.6.38 kernel in tumbleweed.

That is correct. :slight_smile:

No problem with your answer. I just wanted to expand on it. :slight_smile:

However, if Tumbleweed has identical stability of the original distro kernel and same security fixes…why do we still bother with the original repos? I understand the project and (for me as user) I do approve it.

Now I didn’t say “identical stability”. How could that be with “more frequent but stable package upgrades” and “an increase in risk” for example? However if it turns out to be as stable as the standard installation, that will be a bonus. I try to avoid getting into the undefined “average user”. Any user of Tumbleweed will need to carry some additional overhead, and recognize the potential for package management issues arising or bugs if the packagers/authors have different views on stability.

Tumbleweed is new, it attempts to support a more constant and aggressive management of safe updates than conventional updates from the OSS. Note too that Tumbleweed is much more than just kernel updates, it attempts to similarly update a number of other components and Apps.

At least, that’s the way it’s explained (to me). If Tumbleweed proves its value then I assume that it will become the standard rather than the “new feature” it is now.

Tony

The original (OSS) repos (with updates) goes through a lot of testing, … likely it will go through a lot more than a new Tumbleweed update. So I guess one could say there are ‘levels’ of stability, where I believe the OSS repos (with updates) should theoretically be more stable in most cases (but not all) than the Tumbleweed update. Now Tumbleweed will be more stable than factory, and in some (minority) of case even be more stable than OSS repos (with update) where there are major fixes, … but in most cases my guess is Tumbleweed will be the same or less stable than OSS repos.

But note I say guess in cases. The above is just my speculation/guessing and as others on this thread have noted, Tumbleweed is new, and the openSUSE community will need to see how tumbleweed ‘plays out’ before any definitive assessments can be made.

I just installed the 2.6.38 final from the head repo don’t know about the speed but its working…

pawanyadav wrote:
> I just installed the 2.6.38 final from the head repo don’t know about
> the speed but its working…

Rather than using Kernel:Head I believe you’re safer using
Kernel:Stable, which is 2.6.38-1.1 currently.