#2980 install kernel-source fail in genimage in sles11sp2/rh6.3

[BaiYuan]

During genimage in sles11 sp2, there is following error; It happened in redhat6.3 too; And the /install/netboot directory is mounted from other system;
The work around is to umount /install and use local /install
This defect is opened for trace;And it needs restriction;

ERROR:

warning: /install/netboot/sles11.2/x86_64/compute/rootimg/var/cache/zypp/packages/sles11.2/suse/x86_64/iproute2-2.6.29.1-6.22.48.x86_64.rpm: Header V3 RSA/SHA256 signature: NOKEY, key ID 307e3d54
Updating etc/sysconfig/arpd...
Installing: kernel-source-3.0.13-0.27.1 [..error]
Installation of kernel-source-3.0.13-0.27.1 failed:
(with --nodeps --force) Error: Subprocess failed. Error: RPM failed: warning: /install/netboot/sles11.2/x86_64/compute/rootimg/var/cache/zypp/packages/sles11.2/suse/x86_64/kernel-source-3.0.13-0.27.1.x86_64.rpm: Header V3 RSA/SHA256 signature: NOKEY, key ID 307e3d54
installing package kernel-source-3.0.13-0.27.1.x86_64 needs 41717MB on the /install/netboot filesystem
Abort, retry, ignore? [a/r/i] (a): a
Problem occured during or after installation or removal of packages:

[yangsong]

same defect found on rhels6.4-ppc64, opened a bug to LTC
https://bugzilla.linux.ibm.com/show_bug.cgi?id=93202

[yangsong]

according to the feedback from LTC:
"As per the final comment on the Redhat bugzilla, I don't think this bug will be fixed."

I will look for if there is any work around to this defect.

[Lissa Valletta]

Should we give this some priority since it was hit on the mailing list just this week.
Russell Jones - subject
Subject:genimage/yum and /install disk space requirements
i all,

I am working on setting up a CentOS 5 genimage node that will create
CentOS 5 images that will eventually be deployed using a CentOS 6
management node. I am following this guide to do so:
http://sourceforge.net/apps/mediawiki/xcat/index.php?title=Building_a_Stateless_Image_of_a_Different_Architecture_or_OS

I am running into a strange issue with my new genimage node where a
diskless image that is only going to end up being 700 MB in the end is
requesting almost 50GB of space to install:

---

Disk Requirements:
At least 18431MB needed on the /install filesystem.

Never seen this before. Is this a standard thing, or has something gone
awry? Anyone else seen this before?

I have moved the /install to an NFS mount with more space to see if this
satisfies the requirements, but this doesn't seem right....

Xiao Peng referenced this defect with this workaround
The work around is to not use the nfs mounted /install. Or set '--setopt=diskspacecheck=0' for the yum command in genimage.

[yangsong]

As mentioned above, Redhat won't fix the defect in their later distributions, "skipping disk space checking in NFS mounted /install" is a work around. I will try to checkin the workaround into our genimage code before our 2.8.3 candidate build.

[yangsong]

I have checked in the workaround for redhat genimage. on Suse, I cannot find such an option to skip diskspace checking, can we add a limitation for sles in the release notes?

The git log:
2.9:

commit 5e8d0c91e3adfbfaf8fa18ed8515cde8f9e725af
Author: immarvin yangsbj@cn.ibm.comq
Date: Fri Nov 1 04:06:59 2013 -0700

fix defect #2980 install kernel-source fail in genimage in rh6/centos

2.8.3:

commit 9cb3991fece097b1b54a614b78b53d9c5db3d2a2
Author: immarvin yangsbj@cn.ibm.comq
Date: Fri Nov 1 04:09:15 2013 -0700

fix defect #2980 install kernel-source fail in genimage in rh6/centos

[Lissa Valletta]

Note this workaround will need to be added to each release note as the releases are shipped.