After copying the v2 db on all NUC without regenerate the config.yaml file it seems not working correctly but re-generate by the client on all NUC the config.yaml file and this farm re-started to win now every 2-3 days as expected.
The v1 and old version of the windows client break all NUC windows chia app each time a client on the network win. Of course it was necessary to restart all client app each time one win (with older version then 1.3.5)… (this is my older bug of a past month)