But I want to stress my point again: the test you perform is *expected* to be disruptive: we can make libvirtd a bit more robust in terms of overwriting empty apparmor profiles, but should the very same thing happen to instance disks, the data not flushed to HDD will be lost.
Georgy,
I've updated the patch:
https:/ /review. fuel-infra. org/#/c/ 24954
and tested it on a local environment by manually truncating the apparmor profile of a domain first and then starting it via virsh.
I believe, the updated packages here will help:
http:// perestroika- repo-tst. infra.mirantis. net/review/ CR-24954/ mos-repos/ xenial/ master/ pool/main/ libv/libvirt/
^ could you please give it a try?
But I want to stress my point again: the test you perform is *expected* to be disruptive: we can make libvirtd a bit more robust in terms of overwriting empty apparmor profiles, but should the very same thing happen to instance disks, the data not flushed to HDD will be lost.