Suddenly KVM/virt-manager is failing on qcow2 images

[Richard Chan]

Hi, today all my KVMs based on qcow2 images have failed with "not a bootable
hard disk" message.

I am running F13 x64 fully yum updated on AMD Phenom II X4.

All these VMs were working fine (linux,freebsd,winxp).

When I booted a VM from an ISO image and looked at the /dev/vda - there was
a "corrupt" partition table.

When I converted the qcow2 image to raw, and looked at the partition table,
it was fine.

When I switched the VM to use the raw image, it booted and ran fine.

Any known issues with recent KVM/qcow2 interfaction?

Packages:
kernel-2.6.33.6-147.fc13.x86_64
gpxe-roms-qemu-1.0.0-1.fc13.noarch
qemu-common-0.12.3-8.fc13.x86_64
qemu-img-0.12.3-8.fc13.x86_64
qemu-system-x86-0.12.3-8.fc13.x86_64
qemu-kvm-0.12.3-8.fc13.x86_64

Help!? Tks.

[Richard W.M. Jones]

On Sat, Jul 31, 2010 at 12:32:57PM +0800, Richard Chan wrote:
> Hi, today all my KVMs based on qcow2 images have failed with "not a bootable
> hard disk" message.
>
> I am running F13 x64 fully yum updated on AMD Phenom II X4.
>
> All these VMs were working fine (linux,freebsd,winxp).
>
> When I booted a VM from an ISO image and looked at the /dev/vda - there was
> a "corrupt" partition table.

I'd recommend using virt-rescue for this task.

I'm afraid I've not seen the qcow2 problem that you are seeing.
There's no recent qcow2 corruption issue that I'm aware of.  On the
other hand, using qcow2 does increase the opportunity to break things
via user error.  Did you perhaps do any of the following:

- write to a qcow2 backing disk?

- use qcow2 snapshots at all?

- run out of disk space on the host while a VM was writing to qcow2 disk?

- use any host program to write to the disk file?

- have multiple VMs writing to one disk?

- run guestfish / libguestfs in write mode on a live VM?

Also check for regular host disk errors (/var/log/messages, dmesg).

Rich.

[Cole Robinson]

On 07/31/2010 12:32 AM, Richard Chan wrote:
> Hi, today all my KVMs based on qcow2 images have failed with "not a bootable
> hard disk" message.
>
> I am running F13 x64 fully yum updated on AMD Phenom II X4.
>
> All these VMs were working fine (linux,freebsd,winxp).
>
> When I booted a VM from an ISO image and looked at the /dev/vda - there was
> a "corrupt" partition table.
>
> When I converted the qcow2 image to raw, and looked at the partition table,
> it was fine.
>
> When I switched the VM to use the raw image, it booted and ran fine.
>
> Any known issues with recent KVM/qcow2 interfaction?
>
> Packages:
> kernel-2.6.33.6-147.fc13.x86_64
> gpxe-roms-qemu-1.0.0-1.fc13.noarch
> qemu-common-0.12.3-8.fc13.x86_64
> qemu-img-0.12.3-8.fc13.x86_64
> qemu-system-x86-0.12.3-8.fc13.x86_64
> qemu-kvm-0.12.3-8.fc13.x86_64

Are you managing your VMs using libvirt (tools like virsh or
virt-manager?) If so, can you provide virsh dumpxml $vmname and
/var/log/libvirt/qemu/$vmname.log for an affected VM?

- Cole