attached is my current version of a kickstart file for a kde livecd. I've also written a mail to fedora-devel-list with some explanations. I would be really grateful for some comments. :)

I've tried that at first. But the size exploded. So I've used a diff from the package list created by the rpm-version. Maybe a few groups can be inserted - haven't checked that. With this one I was getting the same results compared to older versions of livecd-creator.

On Mon, 2007-03-26 at 16:24 +0200, Sebastian Vahl wrote: > I've tried that at first. But the size exploded. So I've used a diff > from the package list created by the rpm-version. Maybe a few > groups can be inserted - haven't checked that. With this one I was > getting the same results compared to older versions of > livecd-creator. The problem with not doing so is that people are going to not have packages that are considered part of the "base" system... which isn't such a big deal for just the live CD case, but given the installability, we really want to be able to assume that the basic package set being present isn't going to fundamentally differ based on how it was installed.

Am Mon, 26 Mar 2007 10:29:18 -0400 schrieb Jeremy Katz <katzj(a)redhat.com&gt;: > On Mon, 2007-03-26 at 16:24 +0200, Sebastian Vahl wrote: > > I've tried that at first. But the size exploded. So I've used a > > diff from the package list created by the rpm-version. Maybe a > > few groups can be inserted - haven't checked that. With this one > > I was getting the same results compared to older versions of > > livecd-creator. > > The problem with not doing so is that people are going to not have > packages that are considered part of the "base" system... which > isn't such a big deal for just the live CD case, but given the > installability, we really want to be able to assume that the basic > package set being present isn't going to fundamentally differ based > on how it was installed. > You're right. Haven't concerned that in this case. I will also do some testruns with this groups: @core @base @admin-tools @dial-up @hardware-support Sebastian

So I think it would make sense to commit the KDE live cd kickstart file to livecd git and include it in the livecd-tools RPM just like we do for the minimal and GNOME desktop one; thoughts?

Am Mon, 26 Mar 2007 14:23:50 -0400 schrieb David Zeuthen <davidz(a)redhat.com&gt;: > So I think it would make sense to commit the KDE live cd kickstart > file to livecd git and include it in the livecd-tools RPM just like > we do for the minimal and GNOME desktop one; thoughts? > > Another thing is that Sebastian would need to get access to the live > cd repo so he can commit changes to the KDE kickstart file; I'm fine > with that; Sebastian? Me too.

Maybe I'm doing something wrong or the account system needs some extra time today. Attached is my (hopefully) final version for kde test3. Sebastian

Am Mon, 26 Mar 2007 23:50:58 +0200 schrieb Sebastian Vahl <ml(a)deadbabylon.de&gt;: > Maybe I'm doing something wrong or the account system needs some > extra time today. Attached is my (hopefully) final version for kde > test3. > > Sebastian Mhh. maybe not. After installation to harddisk and reboot I'm getting a kernel panic because no init could be found. So it seems that I'm missing an essential package. I will look at this tomorrow. Attached is the full package list. Perhaps someone could point me to the missing package. A few days ago harddisk installation worked fine. Actual spinned gnome-desktop version does not have this error. Sebastian

Am Tue, 27 Mar 2007 11:21:13 +0200 schrieb Sebastian Vahl <ml(a)deadbabylon.de&gt;: > Ok. There is no package missing. But I have no idea what is getting > wrong. If I chroot into the new installed partition (no matter if > from normal installation or from livecd) and reinstall the kernel, > this installation boots then. If I compare the extracted initrds > before and after reinstalling the kernel the only difference is, > that the last one doesn't contain dm-mirror.ko, > dm-mod.ko,dm-snapshot.ko, dm-zero.ko (I'm not using lvm on my > host). And in the init itself sdb1 is replaced with LABEL=/1 in > mkrootdev. > > Is there a way to tell liveinst/anaconda to be very verbose? > --loglevel=debug inserted in /usr/sbin/liveinst does not produce > interesting logs. Here is a "screenshot" from the error. The next point there normally should be to start "Red Hat nash": http://img261.imageshack.us/my.php?image=p3210195gi7.jpg I've also tried a two days old version without the groups in kickstart file. This one does not show the error and that confuses me. Maybe the error is produced by a package that do not exist in old version?

On Tue, 2007-03-27 at 15:47 +0200, Sebastian Vahl wrote: > It seems to be a selinux thing. When installing with selinux enabled > there are no entries in auditd.log. But booting from the new > installation is not possible. When booting the livecd with > enforcing=0 there are still no entries in audit.log but the new > installation could boot. Weird... the KDE disc I spun overnight with your config seems to be okay

aside from being too big for a CD.

> With the first installation I've tried to recreate the initrd > inside a chroot into this partition. when > using /sbin/newkernel-package, setroubleshoot gives me the attached > error. After doing an "restorecon -v /lib/ld-linux.so.2" (inside > chroot) I don't get the error and the new created initrd is working. /sbin/new-kernel-package isn't quite going to do the right thing for the live CD case unfortunately as it won't create the "live initramfs" and instead just generate a "normal" one.

> I will check my hostsystem. The error must be there, I think. Yeah, sounding like it... sorry I don't have any ideas off-hand of what it could be :-/