There are multiple ways to use Hive on Dataproc :
either using Hive jobs API
or Spark SQL.
But for activating it, I just wrote hive as a command in SSH of master node and it started to work. It seems it is already in hadoop enviroment and does not need further activation.