KVM and libvirt common errors and how to fix them

Most KVM problems come down to a handful of messages. This guide lists the KVM and libvirt common errors, what causes them and how to fix each one quickly.

Before you start

• Root access to the KVM host.
• Know where logs live: journalctl -u libvirtd (or the virtqemud service on newer systems) and per-VM logs in /var/log/libvirt/qemu/vmname.log.

"Could not access KVM kernel module" or /dev/kvm missing

• Check CPU flags: grep -E -c '(vmx|svm)' /proc/cpuinfo. If 0, enable VT-x/AMD-V; see enable hardware virtualization.
• Load the module: sudo modprobe kvm_intel or kvm_amd.
• On a VPS this means nested virtualization is not available; see nested virtualization.

"Failed to connect socket to /var/run/libvirt/libvirt-sock"

• The daemon is not running: sudo systemctl enable --now libvirtd (or the modular virtqemud.socket on newer RHEL-family systems).
• Permission issue: add your user to the libvirt group and log in again.

"Network 'default' is not active"

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

If it fails because the network is missing, redefine it from /usr/share/libvirt/networks/default.xml with virsh net-define. A conflict with a host subnet also prevents startup.

"Permission denied" opening a disk image or ISO

• Keep images under /var/lib/libvirt/images.
• Check that every parent directory is executable for the qemu user.
• SELinux: restorecon -Rv /path or label a custom path with semanage fcontext -a -t virt_image_t.
• AppArmor: check dmesg | grep DENIED and adjust the profile.

"Domain already exists" or "operation failed: domain is already active"

The name is taken. Use virsh list --all; remove a stale definition with virsh undefine name (add --nvram for UEFI VMs).

"Cannot get interface MTU on 'br0': No such device"

The bridge does not exist on this host. Create it or change the VM's NIC to an existing bridge. See virtual networking.

"unsupported configuration: guest CPU doesn't match"

Common after migration to different hardware. Use a CPU model both hosts support instead of host-passthrough.

VM starts but shows "No bootable device"

• Wrong boot order, or the disk was created empty and no ISO is attached.
• BIOS/UEFI mismatch with an imported disk.

VM is very slow

• Check that KVM acceleration is used: the VM log should not mention TCG.
• Switch emulated disks/NICs to VirtIO.
• Look for host overcommit: CPU pinning and overcommit.

Frequently asked questions

Where do I see why a VM failed to start?

The error from virsh start plus /var/log/libvirt/qemu/vmname.log.

virsh shows no VMs but they exist

You are connected to qemu:///session; use sudo virsh or -c qemu:///system.

How do I reset a stuck VM?

virsh destroy name forces it off (like pulling the plug), then start it again.

Next steps

• virsh cheat sheet
• Install KVM on Ubuntu
• QEMU guest agent