Organizing Libvirt Hooks for Qemu

Something I have been meaning to write about for a while is my KVM/VFIO-based gaming setup. Yes, I run Linux. And I run Windows, on Linux, in a virtual machine (VM). And it works! I game on it almost every day.

But the longer post has to wait for another day.

Today I am just sharing the short script that lets me keep my Libvirt hooks for Qemu a tad more organized:

1
2
3
4
5
6
7
8

#!/bin/bash

set -e

SCRIPT="./${1}-${2}-${3}"
if [ -f "$SCRIPT" ]; then
    . "$SCRIPT"
fi

You can save the script as /etc/libvirt/hooks/qemu, and then you can put all your VM-specific scripts under /etc/libvirt/hooks/ as separate files:

• {vm}-prepare-begin
• {vm}-start-begin
• {vm}-started-begin
• {vm}-stopped-end
• {vm}-release-end

For example, if you have a win10-prepare-begin script inside the hooks/ directory, it will only be run when you start your win10 VM at the prepare stage.

You can learn more about Libvirt hooks here.

But why not put these scripts inside a qemu.d/ directory? Since v6.5.0, Libvirt lets you put multiple Qemu-specific hooks inside the qemu.d/ directory. Libvirt still runs all the scripts for each of your VM like it runs /etc/libvirt/hooks/qemu.

This post is 27th of my #100DaysToOffload challenge. Want to get involved? Find out more at 100daystooffload.com.