KVM

KVM (Kernel-Kernel‑based Virtual Machine)Machine (KVM) is a hypervisorfull virtualization solution built into the Linux kernel. User space tools such as libvirt provide a standardized way to interface with virtualization engines, not only KVM but also Xen, OpenVZ and VirtualBox. Graphical tools like virt‑manager allow for user-friendly management of virtual machines running on the local machine or on remote hosts running libvirt.

Preparation

KVM supports a wide range of guests, including Linux, BSD and even Windows. It's architecture allows for near-metal performance. To achieve this KVM utilizes hardware assisted virtualization technologies on the host machine's CPU (Intel VT, AMD-V).

On most desktop systems, these virtualization technologies are not enabled out of the box. To check if virtualization technologies are available on your system, use lscpu:

LC_ALL=C.UTF-8 lscpu | grep Virtualization

This will query the CPU's specifications and filter the output for the relevant virtualization feature section.

If the command produces no output, it indicates the system's CPU either does not support hardware assisted virtualization (unlikely if the machine's CPU was purchased in the last 10 years) or the feature is disabled in the machine's firmware options.

Hardware assisted virtualization technologies are a pre-requisite to being able to use KVM. Refer to your mainboard's user manual to learn how to enable the option.

NOTE: Settings related to CPU virtualization features can sometimes be found in the "Overclocking" section of your firmware settings.

Also consider enabling IOMMU (Intel VT-d, AMD-Vi) for direct device pass-through.

NOTE: On Intel-based systems, unless your kernel has the config option CONFIG_INTEL_IOMMU_DEFAULT_ON set (default is unset) you will also have to explicitly add intel_iommu=on to your kernel boot parameters.

Installation

The most common way to start using KVM is by installing QEMU, a generic and open source machine emulator and virtualizer. It utilizes KVM to achieve very good performance and can even emulate a different architecture from the one in the host machine. Though KVM is the most commonly used hypervisor for QEMU, it can also utilize other hypervisors, such as Xen, OpenVZ or VirtualBox.

Arch Linux offers several levels of completeness of the QEMU suite of emulators:

• qemu-full: installs the entirety of QEMU tools and libraries, capable of emulating many different systems and architectures.

Install the package that applies best for your use-case, e.g. if you plan on running virtual machines on your desktop computer:

sudo pacman -S qemu-desktop

QEMU itself does not provide graphical tools to set up and manage virtual machines. This is where libvirt comes in: it provides APIs for user-facing applications to offer graphical front-ends to users. One such application is virt-manager available from Arch repositories:

sudo pacman -S virt-manager libvirt

By default, only root can interface with KVM and thus the full system emulator provided by QEMU and libvirt will require elevated privileges every time you intend to run virtualized environments.

To allow your user to use virtual machines powered by KVM and libvirt add it to the libvirt Packages

yay -S qemu libvirt edk2-ovmf virt-manager nfs-utils virtio-win

# optional dependencies
iptables-nft dnsmasq   # for default NAT/DHCP networking
bridge-utils           # for bridged networking
openbsd-netcat         # for remote management over SSH

sudo usermod -aG libvirt $USER

NOTE:

Then, enable and start the libvirt daemon:

sudo systemctl enable --now libvirtd

Networking

For virtual machines to have network access, a network bridge needs to be created.

This is relatively easily achieved with nmcli, the CLI tool for NetworkManager:

nmcli# connectionCreate addthe typenew bridge ifname br0 con-name "Netzwerkbrücke" stp nointerface
nmcli connection add \
      type bridge \
      ifname br0 \
      con-name "Bridge" \
      stp no

# Determine the current default route network adapter and add it to the bridge
DEFAULT_IF=$(ip r s default | awk '{print $5}')
nmcli connection add \
      type bridge-slave \
      ifname enp39s0"$DEFAULT_IF" \
      con-name "Ethernet" \
      master br0

# Disable the old wired connection and bring up the bridge
nmcli connection down "KabelgebundeneWired VerbindungConnection 1"
nmcli connection up "Netzwerkbrücke"Bridge"

To

nmcli con mod "Netzwerkbrücke" ipv4.method disabled ipv6.method ignore 

<network>
  <name>br0</name>
  <forward mode='bridge'/>
  <bridge name='br0'/>
</network>

Import

sudo virsh -c qemu:///system net-define br0.xml
sudo virsh -c qemu:///system net-autostart br0

Storage

Storage under libvirt is defined as pools which can be any of the following:

The default storage pool is defined as a local directory at /var/lib/libvirt/images. This is where libvirt will store disk images for guests.

sudo chattr +C /var/lib/libvirt/images

Storage pools can be created from within virt-manager or by writing custom XML definitions and importing them with virsh, in the case that the storage pool type is not exposed through the GUI or you need more fine-grained control over the specifications of the pool.

If you have a remote storage poollocation (that holds disk images, i.e.g. remotean exported NFS share with ISO images)images, it's possible to add it as a pool and mount it into the guest without the need to copy the images to your computer first.

The XML definition for a storage pool, named remote-iso.xml for example, could look something like this:

<pool type="netfs">
  <name>iso</name> <!-- name of the pool -->
  <source>
    <host name="dragonhoard"/> <!-- hostname/IP of remote machine -->
    <dir path="/Download/Software/mnt/user/downloads/ISOs"/> <!-- full path to images on remote machine -->
    <format type="auto"/>
  </source>
  <target>
    <path>/var/lib/libvirt/images/iso</path> <!-- mount point on local machine -->
  </target>
</pool>

In

sudo mkdir -p /var/lib/libvirt/images/iso

Finally, import the pool definition and set it to virt-managerautostart:

sudo virsh -c qemu:///system pool-define remote-iso.xml
sudo virsh -c qemu:///system pool-autostart iso

sudo mkdir -p /var/lib/libvirt/images/iso