Comment 5 for bug 1821912

@LIU: please not that in Your 3rd example (http://logs.openstack.org/82/643882/4/check/tempest-slow-py3/3432a15/testr_results.html.gz) it was different (and known) issue. VM got kernel panic. So that case can be easily moved out from our analysis process :)

This finding with FIP in DOWN state is very interesting. I wonder why it is in such state.