1

I have been building VM's on my fedora 37 server for a while, when I suddenly started running out of disc space. So, I looked in the filesystem and saw that the new VM's had their discs built at /var/lib/libvirt/images, so I made /var/lib/libvirt/images a symlink to a new partition which I had created, which has a lot more space. Now, however, when I virt-install a new VM, it gets stuck in the console and never completes the build. I'm not 100% sure the permissions of the new symlink are setup correctly. Here's what I have:

ls -l /var/lib/libvirt
total 0
drwx--x--x. 2 root root   6 Nov 20 12:45 boot
drwxr-xr-x. 2 root root  97 Mar 10 19:33 dnsmasq
drwx--x--x. 2 root root   6 Nov 20 12:45 filesystems
lrwxrwxrwx. 1 root root  15 Mar 20 06:00 images -> /vmroot/images/
drwx------. 2 root root   6 Nov 20 12:45 network
drwxr-x--x. 9 qemu qemu 103 Mar 24 11:32 qemu
drwx--x--x. 2 root root   6 Nov 20 12:45 swtpm

The VM Launch failure looks like this, FWIW:

1;39mmultipathd.[    4.614630] multipathd[555]: /etc/multipath.conf. See man    mpathconf(8) for more details
service…apper [    4.616316] multipathd[555]: /etc/multipath.conf does not exist, blacklisting all devices.
Multipath Device[    4.617884] multipathd[555]: You can run "/sbin/mpathconf --enable" to create
 Controller.[    4.619390] multipathd[555]: /etc/multipath.conf. See man mpathconf(8) for more details

[    4.620934] multipathd[555]: reconfigure: setting up paths and maps
[    4.622211] systemd[1]: Started multipathd.service - Device-Mapper Multipath Device Controller.
[   ***] Job dev-disk-by\x2dlabel-boot.device/start running (20s / 1min 30s)

And, eventually:

[    4.510452] systemd[1]: Started multipathd.service - Device-Mapper Multipath Device Controller.
[[   93.412712] systemd[1]: dev-disk-by\x2dlabel-boot.device: Job dev-disk-by\x2dlabel-boot.device/start timed out.
 TIME ] Timed out waiting fo[   93.418160] systemd[1]: Timed out waiting for device dev-disk-by\x2dlabel-boot.device - /dev/disk/by-label/boot.
r device ▒▒ice - /dev/disk/by-label/boot.

After this, it's stuck in a loop, allowing me to hit Ctrl-D to continue the loop, or Enter for maintenance mode.

As I was able to launch VM's with no trouble before the symlink change, but after I can no longer build VM's, I'm pretty sure this is the problem. But I have no idea how to fix. Any suggestions?

Jason Michael
  • 111
  • 3

1 Answers1

0

This is probably due to selinux.

Try this:

semanage fcontext -a -e /var/lib/libvirt /vmroot/images
restorecon -RF /var/lib/libvirt
restorecon -RF /vmroot/images
systemctl restart libvirtd

If it does not work, you can try to remove the /var/lib/libvirt/images symlink and substituting it with a bind mount doing the following:

semanage fcontext -a -e /var/lib/libvirt /vmroot/images
mkdir /var/lib/libvirt/images
mount --bind /vmroot/images /var/lib/libvirt/images
restorecon -RF /var/lib/libvirt
systemctl restart libvirtd

If it still does not work you can set selinux in permissive mode:

setenforce 0
systemctl restart libvirtd

To permanently set selinux in permissive mode, you need to edit /etc/selinux/config and set

SELINUX=permissive
shodanshok
  • 47,711
  • 7
  • 111
  • 180