I have mostly avoided “btrfs”. But I decided that it was time to give it a fair try.
So I recently installed Leap 15.1 Beta in a KVM virtual machine with a 50G virtual disk. I configured the VM for “virtio” video, since QXL seems to be broken. And I set the install to use UEFI. Beyond that, I mostly took the defaults. So I ended up with “btrfs” for the root file system, and with “/home” as a subvolume of that “btrfs” system. That was perhaps a week ago.
Yesterday, I booted it up, and updated to the lastest release (with “zypper dup”). It now seems to be a release candidate (the “Beta” has disappeared from the name). I installed enigmail and Thunderbird. I uploaded “.thunderbird” from my main desktop. And I checked whether “enigmail” was working (it was).
Then I shutdown. And I noticed a “fail to unmount” message on shutdown. It went too fast to read. So I booted it up again. And the VM was hopelessly slow. It took about 5 minutes for the boot menu to show up. At that point, I forced it off. Several retries, and the same every time. But other VMs started normally, so it wasn’t an obvious failure of virt-manager and associated software.
I connected a recent 15.1 iso as a virtual DVD, and booted that iso. That booted fine. So the virtual machine itself seemed to be working. It was just grub2 from the installed 15.1 that was ridiculously slow.
I booted that iso to the rescue system. Once in the rescue system, I did:
mount /dev/sda2 /mnt ### mount the btrfs root file system
mount --bind /dev /mnt/dev
mount --bind /proc /mnt/proc
mount --bind /sys /mnt/sys
I’ll note that the “btrfs” root file system mounted without any problems.
Continuing, I did:
chroot /mnt
mount -a ### mount everything within the chroot environment
exit ### leave the "chroot" environment
Again, this all went smoothly and quickly. The “df” command showed that all of the subvolumes were properly mounted.
So I rebooted, without having actually done anything other than mounting the “btrfs” file system.
umount -R /mnt
shutdown -r now
The system rebooted just fine. The grub menu from the installed system now worked properly. It was no longer ridiculously slow.
I really don’t know what happened. My guess is that there was a problem with the “btrfs” file system, and mounting then unmounting it was enough to fix it. But that’s only a guess.
That VM is now working just fine again. But it does look as if there’s a “FAIL to unmount /var” on every shutdown.
This leaves me with concerns about “btrfs”. However, if this was a “btrfs” problem, then I’m surprised at how easy it was to fix.
