As mentioned I simplified the joint maintainer scripts to find the trigger of getting "into bad state" Here is the full script which still does gather some data and showcase the container that is dying on restart: http://paste.ubuntu.com/25031379/
To get there was complex, but the TL;DR to get into bad state is "invoke-rc.d libvirtd start" But there are some constraints on ENV and order of commands that I've seen while simplifying this. I'll dive into that after lunch and report back once more later. -- Christian Ehrhardt Software Engineer, Ubuntu Server Canonical Ltd
