I've seen that a couple of folks have gotten Kadischi to build with FC3 and I thought I'd gotten the kinks worked out, but I continue to see this error message at the end of the anaconda run:
making initrd image /tmp/livecd-build_no26/system/lib/modules/None is not a directory.
*** Fatal error: /home/ewhite/src/tools/kadischi/livecd-mkinitrd.sh returned non zero (256) exit code. Aborting execution.
Anaconda can't find a kernel, even though there's a kernel-2.6.9-1.667.i686.rpm in the repository. In fact there are several flavors (i586, smp, etc.) of that kernel -- is that the problem? Anaconda's db search is producing more than one result and reverting back to the initialized value ("None")?
I thought it was b/c I hadn't built my customized repository correct so, instead, pointed Kadischi at my rsync'ed FC3 directory, and used the minimal-livecd ks script provided in the Kadischi distro, producing the same results.
I applied the Kadischi python patched to the anaconda-10.1.0.2-1 installation that's current in the FC3 repository, and the patch process didn't complain; i.e., I'm running on an FC3 system, trying to produce an FC3 Kadischi LiveCD.
Any help is greatly appreciated.
Eric
On Thu, 2006-01-26 at 22:22 +0100, Eric White wrote:
I've seen that a couple of folks have gotten Kadischi to build with FC3 and I thought I'd gotten the kinks worked out, but I continue to see this error message at the end of the anaconda run:
making initrd image /tmp/livecd-build_no26/system/lib/modules/None is not a directory.
*** Fatal error: /home/ewhite/src/tools/kadischi/livecd-mkinitrd.sh returned non zero (256) exit code. Aborting execution.
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=169812
Eric White wrote:
I've seen that a couple of folks have gotten Kadischi to build with FC3 and I thought I'd gotten the kinks worked out, but I continue to see this error message at the end of the anaconda run:
making initrd image /tmp/livecd-build_no26/system/lib/modules/None is not a directory.
*** Fatal error: /home/ewhite/src/tools/kadischi/livecd-mkinitrd.sh returned non zero (256) exit code. Aborting execution.
This patch seems to fix it on the machines here. Along with the above URL, see this also: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=178623
sadly, no joy on this end. After patching, I confirmed the functions.py changes visually and reran kadische against both an FC3 and FC4 (unadulterated) repo that I sync weekly, resulting in the same error.
I've confirmed that the rpm db under livecd-build_no?? contains no installed kernel rpm, which apparently is the source of the problem.
On Thu, 2006-01-26 at 20:08 -0600, J. Hartline wrote:
Eric White wrote:
I've seen that a couple of folks have gotten Kadischi to build with FC3 and I thought I'd gotten the kinks worked out, but I continue to see this error message at the end of the anaconda run:
making initrd image /tmp/livecd-build_no26/system/lib/modules/None is not a directory.
*** Fatal error: /home/ewhite/src/tools/kadischi/livecd-mkinitrd.sh returned non zero (256) exit code. Aborting execution.
This patch seems to fix it on the machines here. Along with the above URL, see this also: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=178623 -- Fedora-livecd-list mailing list Fedora-livecd-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-livecd-list
Eric White wrote:
sadly, no joy on this end. After patching, I confirmed the functions.py changes visually and reran kadische against both an FC3 and FC4 (unadulterated) repo that I sync weekly, resulting in the same error.
I've confirmed that the rpm db under livecd-build_no?? contains no installed kernel rpm, which apparently is the source of the problem.
Could you run over the things you did to create your Fedora/base data files? Also determine all prerequisites are met of the kernel and similar packages by running rpm -qpR <.rpm file> on them, and be sure your repository isn't out of whack.
On Fri, 2006-01-27 at 01:18 -0600, J. Hartline wrote:
Could you run over the things you did to create your Fedora/base data files? Also determine all prerequisites are met of the kernel and similar packages by running rpm -qpR <.rpm file> on them, and be sure your repository isn't out of whack.
I'm using a sync'ed repository from
mirror.web-ster.com::fedora/core/3/i386
so, I'm using the Fedora/base data files that come from there. I've reverted back to a standard repository b/c I wanted to rule out my custom Fedora/base files as the source of the problem.
I've run rpm -qpR <...rpm> and all looks fine, so I'm pretty sure the repository is fine. I've checked my system logs and rsync hasn't complained...
On my FC3 host, I'm using:
busybox-anaconda-1.00.rc1-5 anaconda-runtime-10.1.0.2-1 anaconda-10.1.0.2-1
which I patched from the kadishi distro (without error), even though a later version is noted in the kadischi docs.
I've noticed, however, that when using rpm to inspect my kadischi rpm database, I must specify a fully-qualified path for rpm (e.g., rpm -- dbpath /tmp/livecd-build_no34/system/var/lib/rpm ...) and not a relative path, having cd'd to the livecd system directory.
I've not used an alternate rpm database in the past, so don't know if that's rpm's expected behavior.
Eric
Eric White wrote:
I'm using a sync'ed repository from
mirror.web-ster.com::fedora/core/3/i386
so, I'm using the Fedora/base data files that come from there. I've reverted back to a standard repository b/c I wanted to rule out my custom Fedora/base files as the source of the problem.
On my FC3 host, I'm using:
busybox-anaconda-1.00.rc1-5 anaconda-runtime-10.1.0.2-1 anaconda-10.1.0.2-1
which I patched from the kadishi distro (without error), even though a later version is noted in the kadischi docs.
Eric
Sounds strange, maybe you should try upgrading to a Fedora Core 4 platform and try again. J. Hartline
livecd@lists.fedoraproject.org
-
Eric White -
Ignacio Vazquez-Abrams -
J. Hartline