R H2O with 32-bit java

Question

I am trying to use H2O package in R with 32-bit java. Unfortunately I am restricted by the comapny's IT to install the 64 bit version of java.

How can I make H2O work with 32-bit java, i.e. if possible?

OS - Windows 7

Answer 1

Running H2O with 32-bit Java is really, really not recommended, and unsupported. It will also limit you to very small data sets.

But, if you must, you can run H2O on the command line by hand, and then connect to the running H2O process using the Web UI (or R or Python).

CMD> cd \path\to\h2o
CMD> java -Xmx1g -jar h2o.jar -ip localhost
04-07 15:59:30.483 127.0.0.1:54321       58927  main      INFO: ----- H2O started  -----
04-07 15:59:30.503 127.0.0.1:54321       58927  main      INFO: Build git branch: rel-tutte
04-07 15:59:30.503 127.0.0.1:54321       58927  main      INFO: Build git hash: e27e9002802e76b3ebb56a77f5bb82b827d5b810
04-07 15:59:30.503 127.0.0.1:54321       58927  main      INFO: Build git describe: jenkins-rel-tutte-2
...
04-07 15:59:30.922 127.0.0.1:54321       58927  main      INFO: Registered: 136 REST APIs in: 368mS
04-07 15:59:31.475 127.0.0.1:54321       58927  main      INFO: Registered: 201 schemas in 552ms
04-07 15:59:31.476 127.0.0.1:54321       58927  main      INFO: 
04-07 15:59:31.476 127.0.0.1:54321       58927  main      INFO: Open H2O Flow in your web browser: http://127.0.0.1:54321
04-07 15:59:31.476 127.0.0.1:54321       58927  main      INFO:

Now open your web browser to localhost:54321 as the instruction says at the bottom.

Instead of -Xmx1g you can then experiment with larger heaps to figure out the biggest you can use in your local environment.

[ The "-ip" argument makes H2O's web server bind only to localhost, so you're not accidentally exposing H2O to others not on your computer without realizing it. ]