Hey, Thanks everyone for your replies, I appreciate the feedback! :-) And sorry for the ridiculous delay with which I am answering, I was on vacation. On Tue, 2020-08-04 at 16:21 +0200, Richard Brown wrote:
On Sat, 2020-08-01 at 03:46 +0200, Dario Faggioli wrote:
2) why don't we use an LTSS kernel for Leap?
Dario I'd recommend the following answer every time you get this question
"Because Greg Kroah-Hartman, the #2 Leading Linux Kernel Engineer and the maintainer of the upstream LTS Kernel says that any distribution kernel is better than that LTS Kernel"
Eheh :-)
From talking to Greg on this topic endlessly I know he'd appreciate if we all did a better job of spreading the word and correcting people's expectations about the upstream LTS kernel versions.
His full blog on the topic is here: http://www.kroah.com/log/blog/2018/08/24/what-stable-kernel-should-i-use/
Yes, thanks.
IME, many upstream kernel developers are usually not huge fans of
(enterprise) distros using old kernels and push a lot for using the
latest releases (e.g., Thomas Gleixner often does not loose a single
chance for bushing us about that! :-P).
I know Greg-KH's position (e.g., from that posts). In fact, you often
hear him acknowledging the work of the people in the various distros'
kernel teams (there's one keynote he gave at an OSS where he explicitly
mentions how helpful SUSE kernel engineers where... what was that for,
Meltdown patches for some old kernel, I think?).
All that being said, as far as I've experienced it, the question is
often a little more subtle than "why should I use Leap's kernel instead
than an LTS one?". They're fine with using the Leap kernel, but they
wonder "why Leap doesn't build its kernel on top of an upstream LTS
release?".
In fact, what I think is a little hard to understand --and should
perhaps be communicated better-- is that the work of putting together
and maintaining an enterprise kernel is similar *but not identical* to
the one of maintaining an LTS kernel. For instance, as Greg says in the
post, LTS kernels do get bug and security fixes, but: "no new features
and almost no new hardware support is ever added to these kernels" and
also: "The downsides of using this [LTS] release is that you do not get
the performance improvements that happen in newer kernels".
Instead, we want the SLE/Leap kernel to have some new features and some
performance improvements from the newer kernel (e.g., but not
necessarily only, upon users/customers requests). But that must happen
without updating it entirely, for stability, maintenance, certification
reasons, etc. In fact, we do backport many of those things, in addition
to security and bug fixes. Therefore, having an LTS as a base would
_not_ mean not having to do the work of maintaining the distro's
enterprise kernel, doing backports, etc. Which is indeed what many
users asking the question above seem to think, when they ask it. :-)
So this is pretty much what I try to explain, when I get such
question... Shall anyone have thoughts, corrections, ideas for
improving the argument, I'm all ears! :-)
Thanks again and Regards
--
Dario Faggioli, Ph.D
http://about.me/dario.faggioli
Virtualization Software Engineer
SUSE Labs, SUSE https://www.suse.com/
-------------------------------------------------------------------
<