Setting Up QEMU/KVM on Arch Linux/CachyOS: Virtualization That Doesn't Suck

Why QEMU/KVM?

Here’s the thing: KVM is built into the Linux kernel¹. It’s not a third-party module that might break. QEMU handles the hardware emulation¹, KVM provides near-native performance, and Virt-Manager gives you a GUI that doesn’t make you want to throw your keyboard.

Check Your Hardware First

Before we start, make sure your CPU actually supports virtualization:

lscpu | grep Virtualization

You should see either:

Virtualization: VT-x      # Intel
Virtualization: AMD-V     # AMD

If you see nothing, check your BIOS/UEFI settings. Look for “Intel VT-x”, “AMD-V”, “SVM Mode”, or something similar. It’s usually disabled by default.

Installation

On Arch or CachyOS:

sudo pacman -S qemu-full virt-manager virt-viewer dnsmasq bridge-utils libguestfs swtpm

What you’re getting:

• qemu-full — The actual emulator with all features
• virt-manager — GUI for managing VMs (trust me, you want this)
• virt-viewer — Display client for VMs
• dnsmasq — Handles DHCP/DNS for virtual networks
• bridge-utils — Network bridging tools
• libguestfs — Tools for accessing VM disk images
• swtpm - TPM software emulator built on libtpms.

Enable the Services

The libvirt daemon needs to be running²:

sudo systemctl enable --now libvirtd.service

QEMU backend (for VMs²):

systemctl enable --now libvirtd.socket

For the default NAT network to work automatically:

sudo virsh net-autostart default
sudo virsh net-start default

The thing I had most fun debugging and why I decided to write the blog post is specifying If iptables is to be used or nftables in the configuration:

Important: Uncomment the following in /etc/libvirt/network.conf:

# default: #firewall_backend = "nftables"
firewall_backend = "iptables"

Restart libvirtd.service

sudo systemctl restart libvirtd.service

Add Yourself to the Right Groups

This lets you manage VMs without typing your password every five seconds:

sudo usermod -aG libvirt $USER

Log out and back in for the group changes to take effect. Or if you’re impatient like me:

newgrp libvirt

Using Virt-Manager

Open Virt-Manager from your application menu or run virt-manager. Creating a VM is straightforward:

1. Click the “Create a new virtual machine” button
2. Choose your installation media (ISO file)
3. Allocate CPU and RAM — I usually do 2 cores and 4GB for testing
4. Create a virtual disk or use an existing one
5. Click through and you’re done

Performance Tips

Enable Hugepages (Optional)

For VMs that need serious performance, hugepages reduce memory management overhead:

# Check current hugepages
cat /proc/meminfo | grep Huge

# Enable temporarily (1024 x 2MB = 2GB for VMs)
sudo sysctl -w vm.nr_hugepages=1024

To make it permanent, add to /etc/sysctl.d/40-hugepages.conf:

vm.nr_hugepages = 1024

CPU Pinning

If you’re running a VM that needs consistent performance, you can pin virtual CPUs to physical cores. In Virt-Manager:

1. Open VM details → CPUs
2. Click “Topology” and set sockets/cores/threads manually
3. Under “CPU Pinning”, assign vCPUs to specific host CPUs

Use VirtIO Drivers

When creating VMs, use VirtIO for disk and network. It’s paravirtualized, meaning the guest OS knows it’s in a VM and cooperates instead of pretending to be real hardware.

For Windows guests, you’ll need to load the VirtIO drivers during installation. Grab the ISO from the Fedora project³.

Quick CLI Reference

Sometimes the terminal is faster:

# List all VMs
virsh list --all

# Start a VM
virsh start vm-name

# Graceful shutdown
virsh shutdown vm-name

# Force stop (like pulling the power)
virsh destroy vm-name

# Delete a VM and its storage
virsh undefine vm-name --remove-all-storage

# Take a snapshot
virsh snapshot-create-as vm-name snapshot-name

# Revert to snapshot
virsh snapshot-revert vm-name snapshot-name

Network not available

The default network probably isn’t running:

sudo virsh net-start default

That’s it. You now have a proper virtualization setup that won’t break every time you update your kernel. Welcome to the KVM club.