Skip to content

Latest commit

 

History

History
40 lines (32 loc) · 1.48 KB

Zeppelin.md

File metadata and controls

40 lines (32 loc) · 1.48 KB
Raw

Zeppelin integration

We provide an integration with Apache Zeppelin - a web-bases notebook for data analytics. It allows to write and execute Piglet scripts in notebooks and visualize the results directly. For installation you need

  • the Zeppelin project from here
  • Spark 1.5 built with Scala 2.11
  • the zeppelin branch of Piglet

Setting up the Piglet interpreter requires the following steps:

  1. Build the zeppelin interpreter with sbt:
sbt> package
sbt> assembly
sbt> project zeppelin
sbt> package
  1. Copy the following Jar files to ZEPPELIN_HOME/interpreter/piglet
  • PIGLET_HOME/common/target/scala-2.11/common_2.11-0.3.jar
  • PIGLET_HOME/sparklib/target/scala-2.11/sparklib_2.11-0.3.jar
  • PIGLET_HOME/target/scala-2.11/PigCompiler.jar
  • PIGLET_HOME/zeppelin/target/scala-2.11/piglet-interpreter_2.11-0.3.jar
  • spark-assembly-1.5.2-hadoop2.6.0.jar

  1. Add the Piglet interpreter in ZEPPELIN_HOME/conf/zeppelin-site.xml by adding dbis.piglet.PigletInterpreter to the property value zeppelin.interpreters.

  2. In Zeppelin, go to INTERPRETER and click +Create. Enter a name (piglet) and select the Piglet interpreter from the drop down menu. Currently, no additional properties have to be set.

  3. Create a new notebook, go to "Interpreter binding", and activate piglet %piglet.

  4. Enter your script and mark it as Piglet using %piglet.

  5. Note that you have to use DISPLAY relation instead of DUMP to visualize the result.