Category: KVM

If you are creating a KVM host you can sometimes get the error such XML error: expected unicast mac address. Basically, you if you have manually entered a mac address which meets the Unicast standard then it will cause problems.

Here is a quick way of generating a proper MAC Address:

 openssl rand 6 | xxd -p | sed 's/\(..\)/\1:/g; s/:$//'

I just started to install an Ubuntu 13.10 KVM guest – however the build process failed. The command that I was using was:

vmbuilder kvm ubuntu -v --suite=saucy --libvirt=qemu:///system --arch=amd64  --cpus=1 --mem=2048 --swapsize=2048 --rootsize=220040 --flavour=virtual --hostname=mydomain --ip=1.2.3.4 --mask=1.2.3.4 --net=1.2.3.4 --bcast=1.2.3.4 --gw=1.2.3.4 --dns='8.8.8.8 8.8.4.4' --bridge=br0 --mirror=http://archive.ubuntu.com/ubuntu --components='main,universe' --addpkg=openssh-server --addpkg=acpid --addpkg=acpi-support --user=myuser --pass=mypassword --dest=/mnt/kvm/server/

However, I got this error message:

....
This kernel does not support a non-PAE CPU.
dpkg: error processing /var/cache/apt/archives/linux-image-3.8.0-32-generic_3.8.0-32.47_amd64.deb (--unpack):
 subprocess new pre-installation script returned error exit status 1
Examining /etc/kernel/postrm.d .
run-parts: executing /etc/kernel/postrm.d/initramfs-tools 3.8.0-32-generic /boot/vmlinuz-3.8.0-32-generic
run-parts: executing /etc/kernel/postrm.d/zz-update-grub 3.8.0-32-generic /boot/vmlinuz-3.8.0-32-generic
Errors were encountered while processing:
 /var/cache/apt/archives/linux-image-3.8.0-32-generic_3.8.0-32.47_amd64.deb
E: Sub-process /usr/bin/dpkg returned an error code (1)

Basically, you just need ensure the the linux-image-generic package (i.e. a kernel which supports non-PAE CPU’s) is installed by setting the addpkg flag like so:

--addpkg=linux-image-generic 

I hope this helps someone.

It is very easy to convert a VirtualBox .vdi disk images to KVM qcow2 file. You have to use the RAW file format as an intermediate.

1 – Make sure the VirtualBox machine is shutdown.

2 – Convert the vdi to a raw disk images. Note: vdi are compressed and raw images are not and so you will need to leave enough disk space for entire uncompressed disk.

VBoxManage clonehd --format RAW vm.vdi vm.img

3 – Then on your KVM host:

qemu-img convert -f raw vm.img -O qcow2 vm.qcow2

Hope this is helpful.

I just upgrade Ubuntu on one of my virtual machines and for some reason the file system was corrupted after I rebooted the server. I got errors like:

ext4-fs error ext4_lookup deleted inode referenced

Great. Here is how I fixed the problem.

I found the solution was to attach the qcow2 file as a device on the host and then use fsck. Obviously, you will need to shutdown the virtual machine before doing this.

On the host, enable the nbd (network block device) kernel module. This should be available on all Ubuntu servers after Intrepid.

sudo modprobe nbd max_part=8

You can this use qemu-nbd to connect your qcow2 file as a network block device.

sudo qemu-nbd --connect=/dev/nbd0 /mnt/kvm/wordpress-sites.qcow2

You can then find the partitions using

sudo fdisk /dev/nbd0

On the virtual machine, partition 1 was corrupted, so here is the command I ran:

sudo fsck /dev/nbd0p1

Then disconnect the disk:

sudo qemu-nbd --disconnect /dev/nbd0

Everything seemed to be fine and so I started the domain up again.

I just exported / copied a virtual machine from one of our servers (a HP Proliant) to my laptop so that I could to mess around with it offline. I used the following commands to export the domain.

On the server:

virsh dumpxml wordpress > /tmp/wordpress.xml

On the laptop:

virsh define wordpress.xml

virsh wordpress start

However, I got this error on the laptop:

error: unsupported configuration: guest and host CPU are not compatible: Host CPU does not provide required features: rdtscp, popcnt, sse4.2, dca

The HP server has xeon processors and my laptop has a little core two duo. Obviously the CPU’s had different capabilities. You can see the cpu definitions in the wordpress.xml file were:

less wordpress.xml

<cpu match='exact'>
    <model>Nehalem</model>
    <vendor>Intel</vendor>
    <feature policy='require' name='tm2'/>
    <feature policy='require' name='est'/>
    <feature policy='require' name='monitor'/>
    <feature policy='require' name='ds'/>
    <feature policy='require' name='ss'/>
    <feature policy='require' name='vme'/>
    <feature policy='require' name='rdtscp'/>
    <feature policy='require' name='ht'/>
    <feature policy='require' name='dca'/>
    <feature policy='require' name='pbe'/>
    <feature policy='require' name='tm'/>
    <feature policy='require' name='vmx'/>
    <feature policy='require' name='ds_cpl'/>
    <feature policy='require' name='xtpr'/>
    <feature policy='require' name='acpi'/>
</cpu>

You can also view the same information using virsh capabilities command. It is worth piping it through less because the command produces alot of output and the host cpu information is at the top.

For the laptop:

virsh capabilities | less

<cpu>
    <arch>x86_64</arch>
    <model>Penryn</model>
    <vendor>Intel</vendor>
    <topology sockets='1' cores='2' threads='1'/>
    <feature name='pdcm'/>
    <feature name='xtpr'/>
    <feature name='tm2'/>
    <feature name='est'/>
    <feature name='smx'/>
    <feature name='vmx'/>
    <feature name='ds_cpl'/>
    <feature name='monitor'/>
    <feature name='dtes64'/>
    <feature name='pbe'/>
    <feature name='tm'/>
    <feature name='ht'/>
    <feature name='ss'/>
    <feature name='acpi'/>
    <feature name='ds'/>
    <feature name='vme'/>
</cpu>

So, to fix the issue, I just edited the domains xml definition and put in the correct features.

virsh edit wordpress

Note – if you are using ubuntu on your desktop and have virt-manager installed (which is an awesome gui application which runs on Linux and Macs). You can run virt-manager, open the virtual machine, go to the information tab, select the processor and then “copy host cpu configuration”. I’ve attached a screenshot of how to do this.

I hope this helps someone.