Mailinglist Archive: opensuse-project (244 mails)

< Previous Next >
Re: [opensuse-project] After the Richard talk... what's new for openSUSE
On Tue, 2015-05-05 at 12:59 +0200, oldcpu wrote:
On 05/05/2015 09:34 AM, Michal Kubecek wrote:
On Monday 04 of May 2015 20:56:17 oldcpu wrote:
This 'not too old' kernel has a significant advantage from a support
point of view, which means new hardware support
No, it does not. "Support point of view" is something completely
different: how well are we able to support the kernel packages. And this
would be actually one of the reasons why consider SLE (or SLE based)
kernel even if it may be seen "too old" by some.

Michal Kubeček

Thank you for considering this aspect wrt the kernel.

I suspect we have different interpretations wrt the word 'support', and
I should have been more clear.

When an install fails, or when an installed openSUSE has difficulty with
issues relating to drivers, on the openSUSE forums we often are asked to
help and provide the user 'support' to resolve their problem. Clearly
this is often different from the SLE/openSUSE developer/packager
provision of "support".

Further, I note that there are few regular users (and even less beginner
users) who can patch and custom compile their kernel to incorporate a
feature (with newer hardware support) from a new upstream kernel.
Fortunately, the openSUSE packagers are currently very at finding a good
balance wrt the kernel version selection in each openSUSE version
(reducing the average user requirement for openSUSE kernels with newer
hardware support).

If by openSUSE distribution adopting an older SLE kernel version also
means inclusion of upstream patches (to address new hardware which are
nominally only in a newer kernel) as part of the current nominal
everyday SLE kernel implementation and support approach, then my query
is addressed. Or if Kernel:Stable provides the same , with provision
that the community can help new users install such, and keep such
running (wrt propriatary drivers, virtualization, etc ... ) then that
also addresses my query.

I suspect many of us who monitor the openSUSE mailing list are not
familiar with the details of SLE wrt kernel maintenance and support, and
either a short explanation, or a simple pointer to where we could learn
about such, so to provide better understanding and hence better
contribution, may also be useful.

I can try to clarify things. My team is responsible for working with
IHVs to ensure we have proper support for latest hardware in our
enterprise products. SUSE Linux Enterprise Server/Desktop would not be a
viable solution if customers couldn't install on latest hardware. We
enable new hardware support using a balance between kernel patches that
go out with maintenance updates and updated drivers via extra kernel
module packages. At service pack time we integrate the latest driver
versions as well as more intrusive hardware changes into the kernel.

This allows for the SUSE Linux Enterprise kernels to install and use new
hardware as it's introduced to the market. Certain specific new
features, or optimizations, which are too intrusive to include as a
maintenance update, typically come with the next service pack where we
can integrate and thoroughly QA (together with our partners) reducing
risk of regression to customers running legacy hardware. An important
point to note is that when we do add new hardware support, we work
closely with the our partners for testing to ensure proper functionality
- and when things do break or slip by QA (which does happen) it's a
priority to get fixes out to our customer base. That could be a value to
the openSUSE users.

All that said, because of our core customer base, we do have a focus on
server hardware, and provide the best support there. Desktop/mobile can
have some gaps. Hardware not targeted for enterprise customers hasn't
seen much attention - basically because we are customer demand driven.

If the openSUSE community is interested in leveraging the SUSE Linux
Enterprise kernels, I believe that SUSE would strongly consider and
evaluate any requests for changes in how we build our kernel to close
any gaps to help the community with their needs.

I'm pretty confident that we can find ways of making this "older kernel"
not so much of an issue for hardware support needs of the openSUSE
users. Of course if you want the bleeding edge stuff ASAP, tumbleweed is
the way to go.

< Previous Next >