Dear Maintainer,

virt-install fails with

UnicodeDecodeError: 'utf-8' codec can't decode byte 0xfc in position 10:
invalid start byte

/usr/bin/virt-install --name debug-virt-installer --disk size=20 --cdrom
~/Downloads/debian-13.5.0-amd64-netinst.iso

pretty early on: I guess when it builds a configuration for the VM it is
supposed to create

I did some debugging:
- existing disk image fails in the same way (--import)
- (not) specifying osinfo doesn't change anything
- checked all files it opens (strace --trace file -f) whether byte 10 is 0xfc:
none do.
- the storage pool locations all have ascii paths
- ran with pdb: with pdb, but without debug, the first UnicodeDecodeError
either doesn't occur or is caught.
- ran with LANG=C

I guess this is in either my setup or debian, because I couldn't find bug
report on the internet

Hi,

I had exactly the same issue and finally found the reason in my case:

I was using a volume pool (check with 'virsh pool-list') and for
convenience used my download folder.  After the start of the pool I
must have ('manually') removed some files (disk images?) without
stopping/restarting the pool.

However, this seems to confuse libvirt and the deleted files are still
considered available, leading to the error message.  In my case, it
was completely independet of the installer image, even "virt-install
-d debiantrixie --dry-run" was sufficient to debug.

After stopping and starting the pool again, things worked again as
expected. (virsh pool-destroy Downloads && virsh pool-start Downloads
in my case.)

I downgrade the severity.  Perhaps a more meaningfull error message
is desirable.

Best regards,

  Andi