Getting Started

This section describes how to download and install the Enterprise Steam R client and then create and connect to an H2O cluster.

Downloading and Installing

1. Retrieve the Enterprise Steam R client from the download page.
2. Install Enterprise Steam client for R:

R CMD INSTALL h2osteam_1.7.0.tar.gz

Log in to Enterprise Steam

To log in to Enterprise Steam you will need the URL and your username and password (or access token).

library(h2osteam)
conn <- h2osteam.login(url = "https://steam.example.com:9555",
                       username = "my-username",
                       password = "enter-password-or-access-token-here")

The conn object represents your session, pass it to other methods as a first argument.

Connect to an existing H2O cluster on Hadoop

When you create H2O cluster on Hadoop from the Enterprise Steam Web Client, you can connect to it from R using h2osteam.get_h2o_cluster. You will need the name of the H2O cluster:

cluster_config <- h2osteam.get_h2o_cluster(conn, "my-h2o-cluster")

The cluster_config object is the H2O cluster handle and you can immediately connect to that cluster and begin using H2O:

library(h2o)
h2o.connect(config = cluster_config)

Launching new H2O cluster on Hadoop

You can create a new H2O cluster on Hadoop using the R client:

cluster_config <- h2osteam.launch_h2o_cluster(conn = conn,
                                              name =  "first-cluster-from-R",
                                              profile_name = "default-h2o",
                                              nodes = 2,
                                              node_memory_gb = 30,
                                              version = "3.30.0.1")

After you create a cluster, you can immediately connect to that cluster and begin using H2O:

library(h2o)
h2o.connect(config = cluster_config)

Connect to an existing H2O cluster on Kubernetes

To fetch a cluster running on Kubernetes you can use h2osteam.get_h2o_kubernetes_cluster:

cluster_config <- h2osteam.get_h2o_kubernetes_cluster(conn, "my-h2o-cluster")

The cluster_config object is the H2O cluster handle and you can immediately connect to that cluster and begin using H2O:

library(h2o)
h2o.connect(config = cluster_config)

Launching new H2O cluster on Kubernetes

You can create a new h2o cluster on Kubernetes:

cluster_config <- h2osteam.launch_h2o_kubernetes_cluster(conn = conn,
                                              name =  "first-cluster-from-R",
                                              profile_name = "default-h2o",
                                              nodes = 2,
                                              node_memory_gb = 30,
                                              version = "3.30.0.1")

After you create a cluster, you can immediately connect to that cluster and begin using H2O:

library(h2o)
h2o.connect(config = cluster_config)