Public bug reported:
Binary package hint: python-virtinst
I am trying and testing xen and virt related packages in karmic ubuntu.
Just installed a vm of squeeze via virtinst from the command line in a xen
environment as a xen guest.
The process had two separate short comings.
1. Once the new domain was created, virt-manager listed it, but was unable to
offer a vnc console that connects.
This is a separate problem addressed in the other entry I made, bug 527864,
which is a general description.
2. Once made, the vm is now not able to be accessed due to the following
-- the 'console' made by virt-manager is deficient in its
profile of the new squeeze. Compared with an equivalent made by qemu or
kvm, it offers NOTHING in the boot options of the profile. (i.e. a vm
made with qemu or kvm includes options)
-- the console option itself of the vm while being installed
failed, listing parallel console on a black screen.
-- once installed, the vm predictably failed to boot. By using
virt-manager to open and then run the new vm, it reported
error, boot loader is not executable. (If it had booted, it would
still have failed to present a connected vnc console)
On doing an xml dump of the new vm within virsh, it listed the boot source as
pygrub, but the path to pygrub was just wrong.
A path close to /var/lib/xen-default/boot. Anyway, I used file search and
found pygrub in /usr/bin, not even close to the stipulated path. I made a sym
link so it could be found.
-- next, to seal complete failure, on starting the new vm, it then
reports
xend_post: error from xen daemon:
(xend.err "Boot loader didn't return any data!")
So it's invoked pygrub but doesn't execute a boot. I figure this makes
it another combined slip up between both libvirt0 and python-virtinst,
since it would be libvirt0 creating an xml internal file of the vm with
invalid data. It can start bootup by using kvm to boot it, for which
it's not designed, proving the grub install during installation is ok.
The result is a vm that can't be used. a debian vm created from within
karmic. I may be able to boot it with a guest xen kernel, which would
be a work around, not a fix. I don't expect it to work because it lacks
the content in /lib/modules to support the guest kernel. It was created
with the plan to utilise pygrub and it can't do it. I also can't mount
it via -o loop, which probably can be done but I don't know how to
adjust the required settings.
Need I say I expected the new vm to be accessible via virt-manager with
a full complement of valid settings which effectively booted the new vm.
Overall, karmic is very xen incapable due to this type of error. It's far more
capable under a regular kernel running kvm.
I hoped for better in a current ubuntu.
ide...@karmic:~$ lsb_release -rd
Description: Ubuntu 9.10
Release: 9.10
** Affects: virtinst (Ubuntu)
Importance: Undecided
Status: New
--
virtinst fails to create effective config to boot new vm
https://bugs.launchpad.net/bugs/528384
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
