Comment 19 for bug 1863510

Roman: I think it is, e.g. I think this affects neutron-server/neutron-* restarts which can cause quite a mess.

I also think we should probably do both, raise it again upstream with ansible and implement a workaround on our side. Maybe the upstream isn't likely to change this, though, looking at the old bug report and given the fact that there seems to be a workaround.