Alternatives to preseed installation?

2012-12-06 Thread Christoph Mathys
I'm interested to know about alternatives to preseed installations. As
endproduct, I require an image that I can restore with partclone. The
system is a customized installation of ubuntu-alternate. Our current
process to obtain such an image is as follows:
1. Create a preseed image based on ubuntu-alternate. All modification
need to be part of the preseed process.
2. Install this preseeded iso image inside kvm.
3. Use partclone to create the desired images from the kvm disk image.

I kind of think that it should be possible to eliminate the preseed
iso step. However, I fear that I just end up reimplementing a lot of
stuff that preseed already handles nicely. The annoying part in this
process is installing the iso image. This takes far too long if I need
to iteratively fix errors in the preseed file.

Any buzzwords or tools worth looking at?

Thanks!
Christoph

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss


Re: Alternatives to preseed installation?

2012-12-06 Thread Christoph Mathys
 Maybe FAI?
 http://wiki.fai-project.org/wiki/Installing_Ubuntu_Linux_with_FAI

Thanks for the link. Have you used FAI already? Is it complicated to
learn? Is it reliable?

From a quick look at the website I guess I could also manage
ubuntu/debian servers with this tool.

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss


Re: kernel headers by default revisit...

2012-08-07 Thread Christoph Mathys

I'm not an ubuntu developer, but here's how I see this:

On 07/16/2012 10:04 AM, Daniel J Blueman wrote:

IMHO, the installation and updating of linux-headers-generic,
linux-headers-3.2.0-26, linux-headers-3.2.0-26-generic is unneeded for
~98% of Ubuntu desktop users.


So, 98% are not using nvidia or ati proprietary drivers? Damn, I must 
have a really bad hand when I comes to picking hardware...



Provided the dependencies are correct, it'll be pulled in by eg dkms,
libc-dev etc when expected.


IMHO, there is not one correct here. I'd absolutely hate if packages 
started hardcoding dependencies on linux-headers-generic. They may need 
some linux-headers, but why should that be the ones from generic kernel? 
I'd have to carry a generic headers package whenever I use dkms.


Christoph


--
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss


hang in failsafe.conf on precise

2012-04-24 Thread Christoph Mathys
I just encountered some problems with very long boottimes on precise.
failsafe.conf just hangs until the timeout has elapsed.

The culprit seems to be that I define interfaces in
/etc/network/interfaces that do not exist when I'm testing in kvm
(ifup -a fails). This then seems to prevent static-network-up to be
emitted. I'm not quite sure why this event is never emitted. Is
static-network-up only emitted, if the job networking (exec ifup -a)
runs successfully? (I've disable network-interface.conf)

As a workaround I think I'll just disable failsafe.conf and write my
own job which immediately emits the static-network-up event.

Christoph

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss


Re: ld changed linking behaviour in precise?

2012-04-19 Thread Christoph Mathys
On Thu, Apr 19, 2012 at 2:54 AM, Colin Watson cjwat...@ubuntu.com wrote:
 ... but not compared to precise a few days ago, as Christoph said.  As
 far as I know, --as-needed has been in place for the lifetime of
 precise.

It is of course conveivable that our code changed just enough to
trigger said build failure. I just assumed (because the code worked a
week or so ago) that it probably has something to do with the recent
update to gcc, which invalidated my precompiled headers.

Christoph

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss


Re: ld changed linking behaviour in precise?

2012-04-19 Thread Christoph Mathys
On Wed, Apr 18, 2012 at 4:29 PM, Stefano Rivera stefa...@ubuntu.com wrote:
 Yes, see
 https://wiki.ubuntu.com/OneiricOcelot/ReleaseNotes#GCC_4.6_Toolchain

Very informative, thanks! After reading through the document, I guess
I'll fix the linking instead of turning back on the flag :).

 This is the patch being applied:
 http://bazaar.launchpad.net/~ubuntu-branches/ubuntu/precise/gcc-4.6/precise/view/head:/debian/patches/gcc-as-needed.diff

 It doesn't touch the manpage, that's clearly a bug :)

Will this be taken care of? Should I file a bug report?

Christoph

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss


ld changed linking behaviour in precise?

2012-04-18 Thread Christoph Mathys
Today, my build suddenly had a linker error out of nowhere. Further
investigations revealed:
The default of the linker flag --as-needed seems to be different
compared to lucid and precise a few days ago. While earlier versions
assumed --no-as-needed by default, now the default seems to be to use
--as-needed. Because of that, some libraries which were needlessly
linked against others were not linked anymore. This later caused
'undefined reference' warnings, because said needlessly linked
libraries were no longer recursivly included in other libraries.

Assuming I diagnosed this correctly, was this change intentional?

And for the future: Where can I find the defaults of such command line
switches for gcc/ld? Are there any distro specific overrides? Because
'man ld' still tells me that --no-as-needed is the default...

Christoph

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss


Re: grub2 and Previous Linux versions submenu

2012-04-16 Thread Christoph Mathys
On Fri, Apr 13, 2012 at 1:39 PM, Colin Watson cjwat...@ubuntu.com wrote:
 Indeed; sorry about that.  This patch would do it.

Thank you both. This does exactly what I had in mind. Below an updated
patch that applies cleanly to my current 10_linux:

--- 10_linux.orig   2012-04-16 08:51:36.061262852 +0200
+++ 10_linux2012-04-16 08:52:48.881262843 +0200
@@ -194,7 +194,6 @@
 if [ \$linux_gfx_mode != text ]; then load_video; fi
 EOF

-in_submenu=false
 while [ x$list != x ] ; do
   linux=`version_find_latest $list`
   echo Found linux image: $linux 2
@@ -253,13 +252,4 @@
   fi

   list=`echo $list | tr ' ' '\n' | grep -vx $linux | tr '\n' ' '`
-
-  if [ $list ]  ! $in_submenu; then
-echo submenu \Previous Linux versions\ {
-in_submenu=:
-  fi
 done
-
-if $in_submenu; then
-  echo }
-fi


Christoph

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss


grub2 and Previous Linux versions submenu

2012-04-12 Thread Christoph Mathys
Since some time grub started to hide old kernel entries. This is
usually a good thing. However, on work we routinly boot
older/different kernels and its tedious to always navigate into
Previous Linux versions and choose the correct kernel. Is there a
simple switch that disables the new-style grub menu?

Christoph

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss


Re: boost 1.46 on lucid

2012-01-30 Thread Christoph Mathys
On 01/30/2012 01:08 PM, Kai Mast wrote:
 
 Hey Guys,
 
 I need to backport some of my packages to lucid. They depend on
 boost 1.46. Sadly, I can't find any packages for lucid.

Do you know the tool 'prevu'? I find it extremly handy to do backports
with almost no effort.

For your case: Try and reduce the compat level and debhelper build
dependency and see if it works. If not, you could try backporting
debhelper first...

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss


kdump and self compiled kernel

2012-01-24 Thread Christoph Mathys
I'm trying to get a kernel crashdump from a self-compiled kernel (3.0)
on a 10.04 system. This has been a bit bumpy, but works now with
recent version of kexec-tools, makedumpfile and some hackery on the
script side.

The script 0_kdump expects to find a vmcoreinfo-$KVER. Unfortuanately,
I don't have that file and manual creation failed:
$ makedumpfile -g vmcoreinfo -x
/usr/lib/debug/lib/modules/3.0.17-realtime-1-rt33+/vmlinux
The kernel version is not supported.
The created dumpfile may be incomplete.
generate_vmcoreinfo: Can't find the memory type.

makedumpfile Failed.

Any ideas why I can't create vmcoreinfo from the kernels debug symbols?


Apparently, recent kernels already contain vmcoreinfo in /proc/vmcore.
[0] So I went ahead and changed the invocation of makedumpfile in
'0_kdump' to not pass a reference to the info file (I removed -i
$INFO). This gives me a kernel core dump as expected which I can
analyse with 'crash' and the debug symbols for my kernel.
Could 0_kdump be changed to try -i, -x, and then fall back to not pass
any of the two?


When the system reboots, apport/kernel_crashdump is run, which deletes
the core file and leaves behind .crash file. How can I work with this
file? What information does it contain?

[0] http://lists.infradead.org/pipermail/kexec/2007-August/000521.html

Christoph

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss


recommended way to build upstream kernel for ubuntu

2012-01-06 Thread Christoph Mathys
Which is the recommended way to build my own kernel from kernel.org
source on ubuntu? I need header, kernel+modules and occasionally debug
symbols.

The web is full of documentation about make-kpkg. There are a couple
of wiki pages for ubuntu that document how to rebuild the ubuntu
package for the kernel. I use make-kpkg, which works more or less
after a few lessons learnd, but maybe there is a way of similar
simplicity which just works? I'm also a bit unsure if kernel-package
is still maintained, or if this is a dead end.

Christoph

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss


Re: recommended way to build upstream kernel for ubuntu

2012-01-06 Thread Christoph Mathys
Hi Martin

On Fri, Jan 6, 2012 at 10:27 AM, Martin Pitt martin.p...@ubuntu.com wrote:
 If you just need an upstream kernel, you can use

  http://kernel.ubuntu.com/~kernel-ppa/mainline/daily/

I fear I need a custom configuration and the preempt rt patchset
applied, and the kernel is supposed to run on lucid.

Christoph

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss


Re: apt does not update directory permissions

2011-12-08 Thread Christoph Mathys
 any attempted cure for this would be much worse than the disease.  If
 you want to fix a historical mistake in directory permissions, then yes,
 you need to do so in a maintainer script, and take responsibility for
 any coordination between packages that may need to happen in order to do
 so safely.

Thank you for stating this cristal clear!

Christoph

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss


apt does not update directory permissions

2011-12-07 Thread Christoph Mathys
We are using debian packages to distribute our software inside the
company. Recently I messed up the permissions inside a package: A
whole directory-tree suddenly belonged to root:root, when it should
belong to someone else. The fix for the package was quick, permissions
now look ok again (checking with dpkg -c and clean install). However,
when I update the old (broken) package (dpkg -i or apt-get install,
does not matter), only the file permissions get corrected, all folders
still have wrong permissions.

Is this a feature of apt to only correct file permissions? Do I
seriously need to hack something into postinst to fix this correctly?

Christoph

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss


Re: apt does not update directory permissions

2011-12-07 Thread Christoph Mathys
On 12/07/2011 04:22 PM, Phillip Susi wrote:
 Did you use dpkg-statoverride?

Nope.

-- 
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss


Re: Status of FreeNX team?

2011-06-27 Thread Christoph Mathys

On 05/21/2011 11:25 AM, Maia Kozheva wrote:

As I routinely use NX client and server packages (opennx, freenx, and friends)
as a remote desktop solution, I'm interested in getting them into the official
repository, especially given that existing Ubuntu PPAs are long outdated.


How much active work is going on in this area? The last release of 
freeNX dates back to 2008 and neatx is not developed further.


--
Ubuntu-devel-discuss mailing list
Ubuntu-devel-discuss@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss