KVM host readiness

Prove the host can accelerate guests before debugging libvirt. CPU virtualization, kernel modules, device permissions, memory, and storage all sit below the daemon.

Hardware and kernel proof

lscpu | grep -E 'Virtualization|Hypervisor'
grep -Ewc '(vmx|svm)' /proc/cpuinfo
lsmod | grep -E '^kvm(_intel|_amd)?'
test -c /dev/kvm && stat /dev/kvm

A bare-metal host should expose Intel VT-x (vmx) or AMD-V (svm). Firmware can disable these features even when the processor supports them.

Capacity baseline

free -h
df -hT /var/lib/libvirt
findmnt -no SOURCE,FSTYPE,OPTIONS /var/lib/libvirt

Reserve headroom for the host, page cache, QEMU overhead, and recovery operations. Do not size guests to consume every visible byte.

RHEL-family validation

virt-host-validate qemu

Treat FAIL as a blocker. Review WARN against the intended workload; IOMMU warnings matter for passthrough but not every basic VM host.

If acceleration is unavailable

• Recheck firmware virtualization settings.
• Confirm nested virtualization is exposed when this host is itself a VM.
• Inspect kernel lockdown, module load failures, and /dev/kvm ownership.
• Do not hide the problem by silently falling back to software emulation.