It is reported here:
Downgrading to 3006.1-2 in the fedora template or switching to a debian template for the mgmt VM seems to be the solution for now.