Apache Spark Direct
Connection Type | REST/HTML Server |
Distributions Validated On | Hortonworks 2.6, Cloudera 5.7 |
Server Details | Apache Livy download information can be found here. |
Type of Support | In-Database |
Validated On | Apache Livy 0.3, Apache Spark 1.6, 2.0, 2.1, and 2.2 |
Alteryx Tools Used to Connect
In-database Workflow Processing
Connect to Apache Spark by dragging a Connect In-DB tool or the Apache Spark Code tool onto the canvas. Create a new Livy connection using the Apache Spark Direct driver. Use the instructions below to configure the connection.
Configure the Livy Connection Window
To connect to Livy Server and create an Alteryx connection string:
Add a new In-DB connection, setting DataSource to ApacheSparkDirect. For more information on setting up an In-DB connection, visit Connect In-DB Tool.
On the Read tab, Driver will be locked to Apache Spark Direct. Click the ConnectionString drop-down arrow and select New database connection.
Configure the LivyConnection window.
Livy Server Configuration
Select your security preference:
Enter or paste the Host IP Address or DNS name of the Livy node within your Apache Spark cluster.
Enter the Port used by Livy. The default port is 8998.
Optionally provide the UserName to set user impersonation, the name that Apache Spark will use when running jobs.
Enter or paste the URL of your Knox gateway.
Enter the UserName and Password associated with the specified gateway.
Optionally test the connection:
Select the ApacheSparkVersion used on your cluster.
Select the Kerberos connection type.
Select Test.
Set the ConnectionMode to the coding language to use in the Apache Spark Code tool.
HDFS Connection
Select the ServerConfiguration option that matches the HDFS protocol used to communicate with the cluster.
Enter the Host IP Address or DNS name for the HDFS name node within your Apache Spark cluster.
Enter the Port number. The default port will be populated automatically.
Enter the Host IP Address or DNS name for the HDFS name node within your Apache Spark cluster.
Enter the Port number. The default port will be populated automatically.
Enter or paste the URL of your Knox gateway.
Optionally enter the Username for the HDFS connection.
Optionally enter the Password for the HDFS connection.
Select the Kerberos protocol to use.
Advanced Options
Set the Poll Interval (ms), the time between checks from Alteryx for Apache Spark code execution requests. The default is 1,000 ms, or 1 second.
Set the Wait Time (ms), the time that Alteryx waits for execution requests to complete. Operations that take longer than the set wait time result in a time-out error. The default is 60,000 ms or 1 minute.
The Apache Spark Configuration Options customize the created Apache Spark context and allow advanced users to override the default Apache Spark settings.
Note
By default, the Configuration Option is spark.jars.packages and the Value is com.databricks:spark-csv_2.10:1.5.0,com.databricks:spark-avro_2.10:2.0.1. Depending on your Apache Spark version, you might need to override the default value.
Apache Spark Version | Value |
|---|---|
2.0 - 2.1 | com.databricks:spark-avro_2.11:3.2.0;com.databricks:spark-csv_2.11:1.5.0 |
2.2 | com.databricks:spark-avro_2.11:4.0.0;com.databricks:spark-csv_2.11:1.5.0 |
Select (+ icon) to add another row to the configuration options table.
Select (save icon) to save the current advanced settings as a JSON file. The file can then be loaded into the advanced settings of another connection.
Select (open icon) to load a JSON file into the configuration options table.
Select OK to create your Apache Spark Direct connection.
Limitations
At this time, Alteryx supports native Spark in Cloudera Data Platform (CDP) but not Cloudera Distributed Hadoop (CDH).
TLS/SSL enabled Livy Servers are not supported.