Comment 7 for bug 1641801

Serge said:

> There could well be a libvirt bug as well, but if postinst fails ungracefully
if libvirt is stopped isn't that a packaging bug in uvtool-libvirt?

I disagree. uvtool-libvirt requires libvirtd to be running, and libvirt-bin is expected to ship it running. If a user intervenes by stopping libvirtd or otherwise disabling it, I think the user gets to keep the pieces. Otherwise, what should uvtool-libvirt do gracefully that isn't surprising to the user? The point of a postinst failure is to point out to the user that the package won't work.

John, I think we may have some issue with libvirtd not running as a bug in libvirt packaging. I've seen more than one occurrence of this, but unfortunately we don't have a reliable reproducer.