# Installing and Starting Steam¶

This section describes how administrators can install and start Steam. Refer to the following topics:

## Requirements¶

• Chrome browser with an Internet connection. Note that Chrome is currently the only supported browser.
• Steam tar file
• JDK 1.7 or greater
• H2O jar file for version 3.10.0.7 or greater

## Installing and Starting Steam on YARN¶

A Steam YARN installation provides a method for data scientists and developers to work collaboratively when trainging and deploying models.

### Installation¶

Perform the following steps to install Steam on a YARN edge node.

2. Change directories to the Steam download file, copy the file to your edge node, then untar the Steam file. Be sure to enter the correct password when prompted. For example:

cd ~/Downloads/steam-1.0.0-darwin-amd64
scp -r steam-1.0.0-darwin-amd64 <user>@<yarn_edge_node>:~
tar -xzvf steam-1.0.0-darwin-amd64.tar.gz


### Start Steam on YARN¶

After Steam is installed on the YARN edge node, the next step is to provide the designated Steam admin with the URL of the edge node and the admin login credentials. The admin can then start Steam and begin creating roles, workgroups, and users.

1. SSH into the YARN edge node where the Steam package was copied. Note that this step requires admin privileges.
ssh <user>@<yarn_edge_node>

1. Navigate to the Steam folder and start the Jetty server. This allows you to deploy models and run the Steam Prediction Service.
java -jar var/master/assets/jetty-runner.jar var/master/assets/ROOT.war

1. Open another terminal window and run the following command to start Steam. Be sure to include the --admin-name=admin and --admin-password=admin012 flags. (Or provide a more secure password.) This starts Steam on the edge node at port 9000 and creates a Steam admin. The Steam admin is responsible for creating roles, workgroups, and users and maintains the H2O cluster. Refer to the Steam Start Flags section or use ./steam serve master --help or ./steam serve master -h for information on how to start the compilation and/or prediction service on a different location and for additional flags that can be specified when starting Steam.
sudo ./steam serve master --admin-name=admin --admin-password=admin012


Note: This version of Steam currently includes an experimental/early release of LDAP basic authentication support using the —-authentication-provider and —-authentication-config flags. When used, a configuration file is required. For example:

sudo ./steam serve master --admin-name=admin --admin-password=admin012 —-authentication-provider="ldap-basic" —-authentication-config="file/path.toml"

1. Open a Chrome browser and navigate to the YARN edge node.

## Installing and Starting Steam on a Local Machine¶

Users can download and install steam directly on a local machine without the need for a running instance of YARN.

### Installation¶

2. Change directories to the Steam download file and untar the file.
cd ~/Downloads/steam-1.0.0-darwin-amd64
tar -xzvf steam-1.0.0-darwin-amd64.tar.gz


That’s it! You are now ready to start Steam.

### Start Steam Locally¶

After Steam is installed, the following steps describe how to start Steam.

1. Navigate to the untarred Steam folder.
cd steam--darwin-amd64

1. Start the Jetty server from within the Steam folder. This allows you to deploy models and run the Steam Prediction Service.
java -jar var/master/assets/jetty-runner.jar var/master/assets/ROOT.war

1. Open another terminal window and start Steam. Be sure to include the --admin-name=admin and --admin-password=admin flags. (Or provide a more secure password.) This creates Steam admin. A Steam admin is responsible for creating roles,workgroups, and users. This also starts the Steam web service on localhost:9000, the compilation service on localhost:8080 (same as the Jetty server), and the prediction service on the external IP address of localhost. You can change these using --compilation-service-address=<ip_address:port> and --prediction-service-host=<hostname>. Refer to the Steam Start Flags section or use ./steam serve master --help or ./steam serve master -h to view additional options.
./steam serve master --admin-name=admin --admin-password=admin012


Note: If you are demoing Steam and do not have an Internet connection, you can set the prediction service to point to localhost using --prediction-service-host=localhost.

1. Open a Chrome browser and navigate to http://localhost:9000.

## Steam Start Flags¶

The following table lists the options/flags that can be added to the ./steam serve master command when starting Steam. Use ./steam serve master --help or ./steam serve master -h for the most up-to-date list of available options.

Flag Description
--authentication-config= Specify a configuration file to use for authentication.
--authentication-provider= Specify either basic or digest as the authentication mechanism for client logins.
--cluster-proxy-address= Specify a proxy address. For example: <ip>:<port> or :<port>.
--compilation-service-address= Specify an address to use for the compilation service. For example: <ip>:<port> or :<port>.
--prediction-service-host= Specify the hostname to use for the prediction service.
--prediction-service-port-range= Specify a range of ports to create prediction services on. For example: <from_port>:<to_port>.
--profile= Specify true to enable the Go profiler.
--admin-name= Set the admin username. This is required at first-time-use only.
--admin-password= Set the admin password. This is required at first-time-use only.
--web-address= Specify the web server address. For example: <ip>:<port> or :<port>.
--web-tls-cert-path= Specify the web server TLS certificate path.
--web-tls-key-path= Specify the web server TLK key file path.
--working-directory= Specify the working directory for application files.
--yarn-enable-kerberos= Specify whether to enable Kerberos authentication. This requires a username and keytab.

## Next Steps¶

Now that Steam is up and running, you can log in to the machine that is running Steam and use the CLI to create additional roles, workgroups, and users. Refer to the User Management section.