KVM-Guest does not boot: qemudParsePCIDeviceStrs

Question

I have a Server running Ubuntu 10.10 Server-Edition kvm, and libvirt (both ubuntu-native packages) HDD-Partitioning was done with LVM. Then I created some VMs with Virt-Manager and assigned LVM-Volumes to the VMs.

Now the VMs do not boot. Virt-Manager shows a CPU-Usage of 100% for this Guest and the VNC-Connection states

Booting from Hard Disk

The VM-specific logfiles do not show any abnormality only syslog shows a warning

warning : qemudParsePCIDeviceStrs:1422 : Unexpected exit status '1', qemu probably failed

What can I do to find the error?

score 1 · Answer 1 · answered Jan 10 '11 at 18:37

1

Can you try stopping AppArmor "sudo /etc/init.d/apparmor stop; sudo /etc/init.d/apparmor teardown"

then restarting the VM!

answered Jan 10 '11 at 18:37

kim0

1,130
7
7

this also didn't work. I found out, that creating new VMs is no problem. They run as they should. So I suspect Virt-Manager of having done something wrong with my old VM's. I used Virt-Manager to reorder the Disk-Positions and after that nothing worked anymore. – markus Jan 11 '11 at 11:06

score 0 · Answer 2 · answered Jan 09 '11 at 08:56

0

You might want to try switching the disk type, e.g. from virtio to ide. You can virsh dumpxml > guestname.xml, edit the dump file to make the change, destroy and undefine the guest, then create from the edited dump file.

answered Jan 09 '11 at 08:56

nedm

5,610
5
30
52

As an aside, the guest logfiles have yet to provide me really helpful information when troubleshooting a qemu/kvm problem. This might just be me, but it's one of my few complaints w/ KVM. – nedm Jan 09 '11 at 09:03
I already had the disk type set to SCSI instead of virtio. – markus Jan 09 '11 at 09:07
I'd still switch it to IDE (or even to virtio) to see if it makes a difference. Also, how are the physical disks composing the LVM volumes connected? Are they SATA, running through a PCI/x SATA card by chance? That's one setup I've seen similar (though not identical) errors from. – nedm Jan 09 '11 at 09:14
lspci | grep SATA shows me an Intel onboard SATA-Controller. My current config file: http://pastebin.com/K8DDtqi5 – markus Jan 09 '11 at 09:35
using virtio I get this error: "error: internal error process exited while connecting to monitor: PCI: devfn 24 not available for rtl8139, in use by virtio-blk-pci". using ide does not work at all – markus Jan 09 '11 at 09:43

score 0 · Answer 3 · answered Feb 16 '11 at 15:26

0

Did you ever restart the server after installing libvirt-bin and/or kvm? I have no idea why, but it solved the problem in our case.

answered Feb 16 '11 at 15:26

Michael

285
4
16

Yes I did. But it didn't help. – markus Feb 17 '11 at 10:12
I finally ended up creating a new config with VirtManager with the proper disk-order, setting the disk type to SCSI, reloaded the profile and now it works like a charme. – markus Feb 17 '11 at 10:19

score 0 · Answer 4 · edited Nov 06 '12 at 17:22

I just had exacly same problem after increasing VM memory resources via virt-manager:

warning : qemudParsePCIDeviceStrs:1422 : Unexpected exit status '1'
qemu probably failed

sitting in my /var/log/messages.log after trying to start VM.
I could not start VM. Since:

sudo service libvirt-bin stop
sudo service libvirt-bin start

didn't work on Debian stable 64 I solved this issue with:

ps -ef | grep libvirtd
sudo kill -9 31416
sudo service libvirt-bin start
ps -ef | grep libvirtd

and then:

sudo virsh start SomeOS

I think this is related to inaccesible volume or VM disk image.

KVM-Guest does not boot: qemudParsePCIDeviceStrs

4 Answers4