h2oai / h2o-droplets Goto Github PK

Context: I want to train and serve h2o models from java applications.
I know serving is possibile e.g. via MOJOs, but I'm not sure about training as h2o docs are all about R and Python.
I am aware that I can use h2o flow Rest API from Java, but I'm trying to understand if that's the only option, or whether there are (undocumented) Java APIs for training.

That's how I came across h2o droplets, as they are enlisted in h2o docs about how to run h2o from java.

Thus I'm trying to run the example code, that apparently is the only kind of "guidance" offered.

Problem: the readme simply says to launch grade run, but if I do, nothing "special" happens...I simply end up with a h2o instance running (e.g. I can readily open h2o flow web ui), but e.g. I don't see anything printed on stdout.
Upon code inspection, I suspected the if(){} statement was not executed, so I tried launching gradle run --args="-client", but yet again nothing special happens: the application waits indefinitely for cloud formation (or timeouts, depending on code configuration).
Then I tried launching grade run first, as AFAIK that launches the h2o node, then gradle run --args="-client" after, to connect the client to the node, but this way I get errors on the "server side" (h2o node, not client) about "client connections not accepted"

The lack of documentation hinders any further inquiry...can anybody offer assistance?