Apparently, running a zero node on a different hostname (by moving the zw directory) is not a good idea, as I have learned here.
The log output of the zero node could indicate that situation. Ideally, zero should fail quickly. This situation should be easy to reproduce.