Thanks, Gibi for your analysis, and Ronelle too (for the update in comment#16):
Meanwhile, two points:
(1) The original instance crash that Artom pointed out in comment#2 is not reproducible in later test runs. It seems tobe spurious
(2) The QEMU error: "unable to execute QEMU command 'blockdev-del'" was caused by a bug in libvirt, but that is NOT the root cause of the crash in point (1). This seems to be a latent bug in libvirt, uncovered by this test. And is being fixed upstream (by Peter Krempa):
Thanks, Gibi for your analysis, and Ronelle too (for the update in comment#16):
Meanwhile, two points:
(1) The original instance crash that Artom pointed out in comment#2 is not reproducible in later test runs. It seems tobe spurious
(2) The QEMU error: "unable to execute QEMU command 'blockdev-del'" was caused by a bug in libvirt, but that is NOT the root cause of the crash in point (1). This seems to be a latent bug in libvirt, uncovered by this test. And is being fixed upstream (by Peter Krempa):
https:/ /listman. redhat. com/archives/ libvir- list/2022- February/ msg00790. html
— qemu: blockjob: Avoid spurious log errors when cancelling a shallow copy with reused images