FAQ Search Today's Posts Mark Forums Read
» Video Reviews

» Linux Archive

Linux-archive is a website aiming to archive linux email lists and to make them easily accessible for linux users/developers.


» Sponsor

» Partners

» Sponsor

Go Back   Linux Archive > Debian > Debian Development

 
 
LinkBack Thread Tools
 
Old 05-13-2011, 09:50 PM
 
Default earliest supported kernel is 2.6.32 now

From the udev 168-2 changelog:

* Earliest kernel release supported raised from 2.6.27 to 2.6.32 due
to the usage of accept4(2).


So you may want to clean up your packages to remove all code needed for
compatibility with older kernels.

--
ciao,
Marco
 
Old 05-13-2011, 10:23 PM
Adam Borowski
 
Default earliest supported kernel is 2.6.32 now

On Fri, May 13, 2011 at 11:50:58PM +0200, Marco d'Itri wrote:
> From the udev 168-2 changelog:
>
> * Earliest kernel release supported raised from 2.6.27 to 2.6.32 due
> to the usage of accept4(2).
>
> So you may want to clean up your packages to remove all code needed for
> compatibility with older kernels.

There is a world outside of udev.

Hosting places tend to be really slow at the uptake of new kernels, so
usually you can expect 2.6.26 regardless of whether you run stable, testing
or unstable in your vserver.

Breaking compatibility just for the sake of breaking, or at most premature
cleanup, is unnecessary damage.

Udev is installed only in actual host machines, not in chroots/vservers/
lxc/openvz so not supporting kernels older than stable is acceptable (but
please, never again break compat with stable like squeeze did). This is not
the case for packages which don't touch the hardware directly.

--
1KB // Microsoft corollary to Hanlon's razor:
// Never attribute to stupidity what can be
// adequately explained by malice.
 
Old 05-14-2011, 12:22 AM
Ben Hutchings
 
Default earliest supported kernel is 2.6.32 now

On Sat, May 14, 2011 at 12:23:56AM +0200, Adam Borowski wrote:
> On Fri, May 13, 2011 at 11:50:58PM +0200, Marco d'Itri wrote:
> > From the udev 168-2 changelog:
> >
> > * Earliest kernel release supported raised from 2.6.27 to 2.6.32 due
> > to the usage of accept4(2).
> >
> > So you may want to clean up your packages to remove all code needed for
> > compatibility with older kernels.
>
> There is a world outside of udev.
>
> Hosting places tend to be really slow at the uptake of new kernels, so
> usually you can expect 2.6.26 regardless of whether you run stable, testing
> or unstable in your vserver.
>
> Breaking compatibility just for the sake of breaking, or at most premature
> cleanup, is unnecessary damage.
[...]

Backward-compatibility has a cost, sometimes substantial.

I don't think packages in testing/unstable should be expected to support
any kernel version older than that in stable. It's the same same rule we
apply to any other dependency.

Ben.

--
Ben Hutchings
We get into the habit of living before acquiring the habit of thinking.
- Albert Camus


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 20110514002242.GT2268@decadent.org.uk">http://lists.debian.org/20110514002242.GT2268@decadent.org.uk
 
Old 05-15-2011, 11:14 AM
Russell Coker
 
Default earliest supported kernel is 2.6.32 now

On Sat, 14 May 2011, Ben Hutchings <ben@decadent.org.uk> wrote:
> Backward-compatibility has a cost, sometimes substantial.
>
> I don't think packages in testing/unstable should be expected to support
> any kernel version older than that in stable. It's the same same rule we
> apply to any other dependency.

There is also a cost to running old versions of packages to match the kernel
that you are compelled to use.

EG if you have a RHEL5 system running as a Xen Dom0 it's probably not going to
be a desired upgrade option to use Debian/Squeeze. As RHEL6 doesn't support
Xen Dom0 and RHEL5 doesn't support a Squeeze kernel for the DomU that leaves
no option for a recent Xen kernel with RHEL as Dom0. I'm sure that I'm not
the only person who's running a system with Squeeze DomUs that have a Lenny
udev to deal with this.

--
My Main Blog http://etbe.coker.com.au/
My Documents Blog http://doc.coker.com.au/


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 201105152114.02941.russell@coker.com.au">http://lists.debian.org/201105152114.02941.russell@coker.com.au
 
Old 05-16-2011, 09:10 AM
Ian Campbell
 
Default earliest supported kernel is 2.6.32 now

On Sun, 2011-05-15 at 21:14 +1000, Russell Coker wrote:
> On Sat, 14 May 2011, Ben Hutchings <ben@decadent.org.uk> wrote:
> > Backward-compatibility has a cost, sometimes substantial.
> >
> > I don't think packages in testing/unstable should be expected to support
> > any kernel version older than that in stable. It's the same same rule we
> > apply to any other dependency.
>
> There is also a cost to running old versions of packages to match the kernel
> that you are compelled to use.
>
> EG if you have a RHEL5 system running as a Xen Dom0 it's probably not going to
> be a desired upgrade option to use Debian/Squeeze. As RHEL6 doesn't support
> Xen Dom0 and RHEL5 doesn't support a Squeeze kernel for the DomU that leaves
> no option for a recent Xen kernel with RHEL as Dom0.

You are booting Squeeze using the RHEL5 2.6.18 kernels supplied in /boot
or something like that?

> I'm sure that I'm not
> the only person who's running a system with Squeeze DomUs that have a Lenny
> udev to deal with this.

In general there is no requirement to reuse the dom0 kernel as your domU
kernel, although I appreciate that some hosting providers may add that
sort of requirement (or a similar requirement to use one of a blessed
set of kernels).

If you administer dom0 though you can usually use the actual Squeeze
kernel even for a RHEL5 based dom0. For example by using p[yv]grub, or
alien to install the .deb as a .rpm or even just by extracting the
vmlinuz file from the .deb and dropping it in /usr/local/.

Ian.

--
Ian Campbell
Current Noise: Suffocation - Torn Into Enthrallment

(Presuming for the sake of argument that it's even *possible* to design
better code in Perl than in C. :-)
-- Larry Wall on core code vs. module code design


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 1305537051.31488.214.camel@zakaz.uk.xensource.com" >http://lists.debian.org/1305537051.31488.214.camel@zakaz.uk.xensource.com
 
Old 05-16-2011, 01:42 PM
Russell Coker
 
Default earliest supported kernel is 2.6.32 now

On Mon, 16 May 2011, Ian Campbell <ijc@hellion.org.uk> wrote:
> > There is also a cost to running old versions of packages to match the
> > kernel that you are compelled to use.
> >
> > EG if you have a RHEL5 system running as a Xen Dom0 it's probably not
> > going to be a desired upgrade option to use Debian/Squeeze. As RHEL6
> > doesn't support Xen Dom0 and RHEL5 doesn't support a Squeeze kernel for
> > the DomU that leaves no option for a recent Xen kernel with RHEL as
> > Dom0.
>
> You are booting Squeeze using the RHEL5 2.6.18 kernels supplied in /boot
> or something like that?

Yes, RHEL5 DomU kernel.

> In general there is no requirement to reuse the dom0 kernel as your domU
> kernel, although I appreciate that some hosting providers may add that
> sort of requirement (or a similar requirement to use one of a blessed
> set of kernels).

http://wiki.xensource.com/xenwiki/XenDom0Kernels

The above URL describes the requirements for kernels. A pvops Xen Linux
kernel (EG the one from Squeeze) requires at least Xen 4.0. RHEL5 and CentOS5
have version 3.1.2 of the Xen kernel and thus won't boot a Squeeze kernel.

It may be possible to run a Xen 4.0 kernel with a RHEL5 Dom0 kernel and user-
space, but that would break the Red Hat support contract and confuse the
support people at the DC.

> If you administer dom0 though you can usually use the actual Squeeze
> kernel even for a RHEL5 based dom0. For example by using p[yv]grub, or
> alien to install the .deb as a .rpm or even just by extracting the
> vmlinuz file from the .deb and dropping it in /usr/local/.

I've tried that and it didn't recognise the kernel. Nothing worked with a
Squeeze kernel without getting a Squeeze Xen kernel.

If you have it working then please tell us the exact versions of the Xen,
RHEL, and Debian kernels in question as well as the architecture it was
running on.

--
My Main Blog http://etbe.coker.com.au/
My Documents Blog http://doc.coker.com.au/


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 201105162342.52339.russell@coker.com.au">http://lists.debian.org/201105162342.52339.russell@coker.com.au
 
Old 05-16-2011, 02:03 PM
Bastian Blank
 
Default earliest supported kernel is 2.6.32 now

On Mon, May 16, 2011 at 11:42:52PM +1000, Russell Coker wrote:
> > In general there is no requirement to reuse the dom0 kernel as your domU
> > kernel, although I appreciate that some hosting providers may add that
> > sort of requirement (or a similar requirement to use one of a blessed
> > set of kernels).
> http://wiki.xensource.com/xenwiki/XenDom0Kernels
> The above URL describes the requirements for kernels. A pvops Xen Linux
> kernel (EG the one from Squeeze) requires at least Xen 4.0.

Please stop spreading fud. This information aplies to Dom0 usage
only[1]. The kernel runs as DomU at least since 3.3 and the patched
Lenny 3.2[2].

> RHEL5 and CentOS5
> have version 3.1.2 of the Xen kernel and thus won't boot a Squeeze kernel.

And the error may be what?

Bastian

[1]: This version got a new interrupt handling.
[2]: #474509
--
I'm a soldier, not a diplomat. I can only tell the truth.
-- Kirk, "Errand of Mercy", stardate 3198.9


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 20110516140329.GA18963@wavehammer.waldi.eu.org">ht tp://lists.debian.org/20110516140329.GA18963@wavehammer.waldi.eu.org
 
Old 05-16-2011, 02:22 PM
Ian Campbell
 
Default earliest supported kernel is 2.6.32 now

On Mon, 2011-05-16 at 23:42 +1000, Russell Coker wrote:
> On Mon, 16 May 2011, Ian Campbell <ijc@hellion.org.uk> wrote:
> > > There is also a cost to running old versions of packages to match the
> > > kernel that you are compelled to use.
> > >
> > > EG if you have a RHEL5 system running as a Xen Dom0 it's probably not
> > > going to be a desired upgrade option to use Debian/Squeeze. As RHEL6
> > > doesn't support Xen Dom0 and RHEL5 doesn't support a Squeeze kernel for
> > > the DomU that leaves no option for a recent Xen kernel with RHEL as
> > > Dom0.
> >
> > You are booting Squeeze using the RHEL5 2.6.18 kernels supplied in /boot
> > or something like that?
>
> Yes, RHEL5 DomU kernel.
>
> > In general there is no requirement to reuse the dom0 kernel as your domU
> > kernel, although I appreciate that some hosting providers may add that
> > sort of requirement (or a similar requirement to use one of a blessed
> > set of kernels).
>
> http://wiki.xensource.com/xenwiki/XenDom0Kernels
>
> The above URL describes the requirements for kernels.

As Bastian points out this is for dom0 only. For domU there is
http://wiki.xensource.com/xenwiki/XenDomUSupport

> > If you administer dom0 though you can usually use the actual Squeeze
> > kernel even for a RHEL5 based dom0. For example by using p[yv]grub, or
> > alien to install the .deb as a .rpm or even just by extracting the
> > vmlinuz file from the .deb and dropping it in /usr/local/.
>
> I've tried that and it didn't recognise the kernel. Nothing worked with a
> Squeeze kernel without getting a Squeeze Xen kernel.
>
> If you have it working then please tell us the exact versions of the Xen,
> RHEL, and Debian kernels in question as well as the architecture it was
> running on.

It's more a general principal that this sort of thing _should_ work and
if it doesn't we would like to hear about the specifics.

Red Hat's own documentation describes how to install PV RHEL 6 (which
uses a 2.6.32 kernel like Squeeze does) on a RHEL 5 host:

http://docs.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/6/html/Virtualization/sect-Virtualization-Installing_Red_Hat_Enterprise_Linux_5_as_a_para_vi rtualized_guest.html

Ian.
--
Ian Campbell
Current Noise: Mono - Burial At Sea

It is impossible for an optimist to be pleasantly surprised.


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 1305555777.20907.24.camel@zakaz.uk.xensource.com"> http://lists.debian.org/1305555777.20907.24.camel@zakaz.uk.xensource.com
 
Old 05-16-2011, 02:58 PM
Russell Coker
 
Default earliest supported kernel is 2.6.32 now

On Tue, 17 May 2011, Bastian Blank <waldi@debian.org> wrote:
> On Mon, May 16, 2011 at 11:42:52PM +1000, Russell Coker wrote:
> > > In general there is no requirement to reuse the dom0 kernel as your
> > > domU kernel, although I appreciate that some hosting providers may add
> > > that sort of requirement (or a similar requirement to use one of a
> > > blessed set of kernels).
> >
> > http://wiki.xensource.com/xenwiki/XenDom0Kernels
> > The above URL describes the requirements for kernels. A pvops Xen Linux
> > kernel (EG the one from Squeeze) requires at least Xen 4.0.
>
> Please stop spreading fud.

http://en.wikipedia.org/wiki/Fear,_uncertainty_and_doubt

Please read the above Wikipedia page. You will find nothing in that page to
describe observing a product to not work, reading documentation that appears
to explain why it doesn't work, and telling others.

> > have version 3.1.2 of the Xen kernel and thus won't boot a Squeeze
> > kernel.
>
> And the error may be what?

I've just tested it again and it works. I didn't make a record of what the
error was, Xen just didn't seem to like the kernel and it aborted early in the
DomU creation process.

Back to the main topic, there are always reasons for not using a particular
kernel. RHEL5 had Ext4 support when Lenny had Ext4dev, I guess I could have
enabled Ext4dev on my Lenny systems that needed something better than Ext3 but
it seemed like a safer bet to use the RHEL5 kernel.

Finally sometimes kernels just crash for no reason that's easy to debug. Sure
it would be nice to be able to devote a lot of time to determining the cause
and making a good bug report, but when a system that has been running reliably
for years fails after an upgrade it's a lot easier to just keep using the old
kernel.

We support things like running an AMD64 kernel with i386 user-space. It would
be nice if we could support a Lenny kernel with Squeeze user-space for as long
as Lenny is supported.

--
My Main Blog http://etbe.coker.com.au/
My Documents Blog http://doc.coker.com.au/


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 201105170058.03065.russell@coker.com.au">http://lists.debian.org/201105170058.03065.russell@coker.com.au
 

Thread Tools




All times are GMT. The time now is 11:02 AM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.
Copyright 2007 - 2008, www.linux-archive.org