Skip to content

Apache Zeppelin – build from source

Apache Zeppelin – build from source

Zeppelin is a notebook based framework for data analysis and data visualizations.
Although building it from the source is quite straightforward I had a few configuration issues that made the process longer.
I used CDH 5.8.0. VM.

Contents

TROUBLE SHOOTING

1. ERROR:

SOLUTION:
npm is not installed. To solve this issue just install npm on your node. i.e with yum:

 

2. ERROR:

SOLUTION:
Two things may occur here. First – jar file can’t be found in the specified repo. Try changing it to the default one by skipping the -Pvendor-repo option and run failing installation step again without it.
Other cause may be that you try to use outdated or not existing zeppelin-zengine.jar. Change the version to the most recent one in your pom.xml file. For reference go to Maven repository.

In my case:

 

3. ERROR:

SOLUTION:
Specified zeppelin-zengine.jar can’t be found. It’s either outdated or doesn’t exist. Check up the most recent one in Maven repository and update your pom.xml file for Zeppelin Server.

In my case:

 

4. ERROR:

SOLUTION:
To skip this error run command:

and rerun failing step once again.

 

5. ERROR:

SOLUTION:
In this case you need to manually update packaged versions of zeppelin-server.jar and zeppelin-web.war in your zeppelin-distribution pom.xml file.

In my case:

 

HOW TO

How to build Zeppelin from source with Maven 3.3.9 from the command line:

1. Get the source from the repository and clone it into desired folder:

Note: Make sure that your Maven version is at least 3.1.0.

2. Build Zeppelin with Maven:

I was building with Spark 1.6 and Hadoop 2.6. You may of course add some other features like SparkR or PySpark support (for more info refer to the official webpage).

3. Start Zeppelin daemons:

By default it runs on http://localhost:8080.

Leave a Reply

Your email address will not be published. Required fields are marked *