On Sat, Jul 6, 2013 at 2:27 AM, Russell Jones <russell@xxxxxxxxxxxx> wrote:
This is installing into a separate install root (via --installroot), not the base system itself. This is then later used to create diskless PXE boot images for booting client systems.
This works fine on CentOS 6.3, but as you can see generates that error on CentOS 6.4. Trying to get the node used to create these images up to CentOS 6.4 if I can.
Short version: Don't.
Long version:
There are enough differences in Berkeley db versions used in RHEL/Centos 5.x and 6.x in that even if your comand succeeded, you will NOT be able to correctly run rpm/yum from INSIDE the newly-created 5.x images.
There's a workaround though:
- run your command once, or create a simple 5.x image which only contains yum
- chroot inside that image (even if you have the error you mentioned), and re-run the command to create a second 5.x image. This time it's run using 5.x's version of yum and bdb, so you should not get the error.
- move the second image outside of the first one
- delete the first one
An example of how that's done is on https://github.com/fajarnugraha/lxc/blob/centos-template/templates/lxc-centos.in (search "download_centos"). It does "yum --installroot install", then check two things:
- is rpmdb located under $HOME (as is the case when run under debian/ubuntu). If yes, then move it to /var/lib/rpm (the correct place for RHEL/Centos)
- does rpm command works inside the new image (checked with rpm -q yum). If not, then re-run the "yum install" command from inside the image
The template was tested to create Centos 5 and 6 images on latest Ubuntu (which should have similar rpm/yum/bdb version as centos 6.x), and can create good centos 5.x and 6.x images.
--
Fajar
_______________________________________________ Yum mailing list Yum@xxxxxxxxxxxxxxxxx http://lists.baseurl.org/mailman/listinfo/yum
