1
votes

This is a totally newbie question to see if I am missing something key (like there is more to install?).

After installing H20 (python 2.7) on a 9 node Hadoop / Spark cluster using pip install of the whl file (h2o-3.10.4.8-py2.py3-none-any.whl) (which says it installed correctly).....

I can import h2o successfully.

But, when I run h2o.init() then I get:

"Checking whether there is an H20 instance running at http://localhost:54321. connected."

But then an error is thrown:

H2oServerError: HTTP 500 Server Error: u'Error: 500'

Should I be able to run H20 by simply pip installing that whl or is there more? The documentation seems outdated and there are lots of different versions found online. Anyone have any experience with this?

1
The docs are versioned by release, but the up-to-date version can always be found at docs.h2o.ai.Erin LeDell
I was thinking more along lines of the various tutorials. So, is there more to do to install the software than what I did? Id also like to install pysparkling.B_Miner
@B_Miner could you provide more details how are you using Python client. Normally, you use h2o driver to start cluster on Hadoop and then connect with help of Python client from any machine (e.g., your laptop). Moreover, if you have logs with exception, they will be helpfulMichal

1 Answers

1
votes

It's most likely that you got this problem solved, but maybe someone else may benefit. Use the install in Python Tab on the following website : http://h2o-release.s3.amazonaws.com/h2o/rel-tutte/2/index.html.