KVM

KVM is configured as the default hypervisor for Compute.

	Note
	There are several sections about hypervisor selection in this document. If you are reading this document linearly, you do not want to load the KVM module prior to installing nova-compute. The nova-compute service depends on qemu-kvm which installs `/lib/udev/rules.d/45-qemu-kvm.rules`, which sets the correct permissions on the /dev/kvm device node.

To enable KVM explicitly, add the following configuration options /etc/nova/nova.conf:

compute_driver=libvirt.LibvirtDriver
libvirt_type=kvm

The KVM hypervisor supports the following virtual machine image formats:

Raw
QEMU Copy-on-write (qcow2)
VMWare virtual machine disk format (vmdk)

The rest of this section describes how to enable KVM on your system. You may also wish to consult distribution-specific documentation:

Fedora: Getting started with virtualization from the Fedora project wiki.
Ubuntu: KVM/Installation from the Community Ubuntu documentation.
Debian: Virtualization with KVM from the Debian handbook.
RHEL: Installing virtualization packages on an existing Red Hat Enterprise Linux system from the Red Hat Enterprise Linux Virtualization Host Configuration and Guest Installation Guide.
openSUSE: Installing KVM from the openSUSE Virtualization with KVM manual.
SLES: Installing KVM from the SUSE Linux Enterprise Server Virtualization with KVM manual.

Checking for hardware virtualization support

The processors of your compute host need to support virtualization technology (VT) to use KVM.

If you are running on Ubuntu, install the cpu package and use the kvm-ok command to check if your processor has VT support, it is enabled in the BIOS, and KVM is installed properly, as root:

                # apt-get install cpu
                # kvm-ok

If KVM is enabled, the output should look something like:

INFO: /dev/kvm exists
KVM acceleration can be used

If KVM is not enabled, the output should look something like:

INFO: Your CPU does not support KVM extensions
KVM acceleration can NOT be used

In the case that KVM acceleration is not supported, Compute should be configured to use a different hypervisor, such as QEMU or Xen.

On distributions that don't have kvm-ok, you can check if your processor has VT support by looking at the processor flags in the /proc/cpuinfo file. For Intel processors, look for the vmx flag, and for AMD processors, look for the svm flag. A simple way to check is to run the following command and see if there is any output:

$ egrep '(vmx|svm)' --color=always /proc/cpuinfo

Some systems require that you enable VT support in the system BIOS. If you believe your processor supports hardware acceleration but the above command produced no output, you may need to reboot your machine, enter the system BIOS, and enable the VT option.

Enabling KVM

KVM requires the kvm and either kvm-intel or kvm-amd modules to be loaded. This may have been configured automatically on your distribution when KVM is installed.

You can check that they have been loaded using lsmod, as follows, with expected output for Intel-based processors:

$ lsmod | grep kvm
kvm_intel             137721  9
kvm                   415459  1 kvm_intel

The following sections describe how to load the kernel modules for Intel-based and AMD-based processors if they were not loaded automatically by your distribution's KVM installation process.

Intel-based processors

If your compute host is Intel-based, run the following as root to load the kernel modules:

# modprobe kvm
# modprobe kvm-intel

Add the following lines to /etc/modules so that these modules will load on reboot:

kvm
kvm-intel

AMD-based processors

If your compute host is AMD-based, run the following as root to load the kernel modules:

# modprobe kvm
# modprobe kvm-amd

Add the following lines to /etc/modules so that these modules will load on reboot:

kvm
kvm-amd

Specifying the CPU model of KVM guests

The Compute service allows you to control the guest CPU model that is exposed to KVM virtual machines. Use cases include:

To maximise performance of virtual machines by exposing new host CPU features to the guest
To ensure a consistent default CPU across all machines, removing reliance of variable QEMU defaults

In libvirt, the CPU is specified by providing a base CPU model name (which is a shorthand for a set of feature flags), a set of additional feature flags, and the topology (sockets/cores/threads). The libvirt KVM driver provides a number of standard CPU model names. Examples of model names include:

"486", "pentium", "pentium2", "pentiumpro", "coreduo", "n270", "pentiumpro", "qemu32", "kvm32", "cpu64-rhel5", "cpu64-rhel5", "kvm64", "pentiumpro", "Conroe" "Penryn", "Nehalem", "Westmere", "pentiumpro", "cpu64-rhel5", "cpu64-rhel5", "Opteron_G1", "Opteron_G2", "Opteron_G3", "Opteron_G4"

These models are defined in the file /usr/share/libvirt/cpu_map.xml. Check this file to determine which models are supported by your local installation.

There are two Compute configuration options that determine the type of CPU model exposed to the hypervisor when using KVM, libvirt_cpu_mode and libvirt_cpu_model.

The libvirt_cpu_mode option can take one of four values: none, host-passthrough, host-model and custom.

Host model (default for KVM & QEMU)

If your nova.conf contains libvirt_cpu_mode=host-model, libvirt will identify the CPU model in /usr/share/libvirt/cpu_map.xml which most closely matches the host, and then request additional CPU flags to complete the match. This should give close to maximum functionality/performance, which maintaining good reliability/compatibility if the guest is migrated to another host with slightly different host CPUs.

Host passthrough

If your nova.conf contains libvirt_cpu_mode=host-passthrough, libvirt will tell KVM to passthrough the host CPU with no modifications. The difference to host-model, instead of just matching feature flags, every last detail of the host CPU is matched. This gives absolutely best performance, and can be important to some apps which check low level CPU details, but it comes at a cost with respect to migration: the guest can only be migrated to an exactly matching host CPU.

Custom

If your nova.conf contains libvirt_cpu_mode=custom, you can explicitly specify one of the supported named model using the libvirt_cpu_model configuration option. For example, to configure the KVM guests to expose Nehalem CPUs, your nova.conf should contain:

libvirt_cpu_mode=custom
libvirt_cpu_model=Nehalem

None (default for all libvirt-driven hypervisors other than KVM & QEMU)

If your nova.conf contains libvirt_cpu_mode=none, then libvirt will not specify any CPU model at all. It will leave it up to the hypervisor to choose the default model. This setting is equivalent to the Compute service behavior prior to the Folsom release.

Troubleshooting

Trying to launch a new virtual machine instance fails with the ERROR state, and the following error appears in /var/log/nova/nova-compute.log

libvirtError: internal error no supported architecture for os type 'hvm'

This is a symptom that the KVM kernel modules have not been loaded.

If you cannot start VMs after installation without rebooting, it's possible the permissions are not correct. This can happen if you load the KVM module before you've installed nova-compute. To check the permissions, run ls -l /dev/kvm to see whether the group is set to kvm. If not, run sudo udevadm trigger.

Legal notices