This instead is an issue of your env:
your additional host is still named 'localhost.localdomain': of course
the engine VM will be able to resolve it but it will not reach your
host; in that case the engine will just try to add the engine VM
itself as an host creating a mess.
Since we saw this issue many times in the past from user logs, we are
now explicitly rejecting 'localhost.localdomain' as an host address.
This is that part that confused me a bit : I need to set up a new FQDN for the engine, but
a different one right ? So each engine on each node needs a different FQDN ?