How to Build Zeppelin from Source

0. Requirements

If you want to build from source, you must first install the following dependencies:

Name Value
Git (Any Version)
Maven 3.6.3 or higher
OpenJDK or Oracle JDK 1.8 (151+)

If you haven't installed Git and Maven yet, check the Build requirements section and follow the step by step instructions from there.

1. Clone the Apache Zeppelin repository

git clone

2. Build source

You can build Zeppelin with following maven command:

./mvnw clean package -DskipTests [Options]

Check build-profiles section for further build options. If you are behind proxy, follow instructions in Proxy setting section.

If you're interested in contribution, please check Contributing to Apache Zeppelin (Code) and Contributing to Apache Zeppelin (Website).

3. Done

You can directly start Zeppelin by running the following command after successful build:

./bin/ start

Build profiles

Spark Interpreter

To be noticed, the spark profiles here only affect the unit test (no need to specify SPARK_HOME) of spark interpreter. Zeppelin doesn't require you to build with different spark to make different versions of spark work in Zeppelin. You can run different versions of Spark in Zeppelin as long as you specify SPARK_HOME. Actually Zeppelin supports all the versions of Spark from 3.2 to 3.5.

To build with a specific Spark version or scala versions, define one or more of the following profiles and options:


Set spark major version

Available profiles are


minor version can be adjusted by -Dspark.version=x.x.x

-Pspark-scala-[version] (optional)

To be noticed, these profiles also only affect the unit test (no need to specify SPARK_HOME) of Spark interpreter. Actually Zeppelin supports all the versions of scala (2.12, 2.13) in Spark interpreter as long as you specify SPARK_HOME.

Available profiles are


Build hadoop with Zeppelin (-Phadoop[version])

To be noticed, hadoop profiles only affect Zeppelin server, it doesn't affect any interpreter. Zeppelin server use hadoop in some cases, such as using hdfs as notebook storage. You can check this page for more details about how to configure hadoop in Zeppelin.

Set hadoop major version (default hadoop3). Available profiles are


minor version can be adjusted by -Dhadoop.version=x.x.x

-Pvendor-repo (optional)

enable 3rd party vendor repository (Cloudera, Hortonworks)

-Pexamples (optional)

Build examples under zeppelin-examples directory

Build command examples

Here are some examples with several options:

# build with spark-3.3, spark-scala-2.12
./mvnw clean package -Pspark-3.3 -Pspark-scala-2.12 -DskipTests

# build with spark-3.4, spark-scala-2.13
./mvnw clean package -Pspark-3.4 -Pspark-scala-2.13 -DskipTests

Ignite Interpreter

./mvnw clean package -Dignite.version=1.9.0 -DskipTests

Optional configurations

Here are additional configurations that could be optionally tuned using the trailing -D option for maven commands

Spark package

spark.archive # default spark-${spark.version} # default${spark.archive}.tgz # default${spark.archive}-bin-without-hadoop.tgz

Py4J package

python.py4j.version # default
pypi.repo.url # default
python.py4j.repo.folder # default /64/5c/01e13b68e8caafece40d549f232c9b5677ad1016071a48d04cc3895acaa3

final URL location for Py4J package will be produced as following:


Frontend Maven Plugin configurations

plugin.frontend.nodeDownloadRoot # default
plugin.frontend.npmDownloadRoot # default
plugin.frontend.yarnDownloadRoot # default

Build requirements

Install requirements

If you don't have requirements prepared, install it. (The installation method may vary according to your environment, example is for Ubuntu.)

sudo apt-get update
sudo apt-get install git
sudo apt-get install openjdk-8-jdk
sudo apt-get install npm
sudo apt-get install libfontconfig
sudo apt-get install r-base-dev
sudo apt-get install r-cran-evaluate

Notes: - Ensure node is installed by running node --version
- Ensure maven is running version 3.6.3 or higher with ./mvnw -version - Configure maven to use more memory than usual by export MAVEN_OPTS="-Xmx2g -XX:MaxMetaspaceSize=512m"

Proxy setting (optional)

If you're behind the proxy, you'll need to configure maven and npm to pass through it.

First of all, configure maven in your ~/.m2/settings.xml.

      <!-- <username>usr</username>
      <password>pwd</password> -->
      <!-- <username>usr</username>
      <password>pwd</password> -->

Then, next commands will configure npm.

npm config set proxy http://localhost:3128
npm config set https-proxy http://localhost:3128
npm config set registry ""
npm config set strict-ssl false

Configure git as well

git config --global http.proxy http://localhost:3128
git config --global https.proxy http://localhost:3128
git config --global url."http://".insteadOf git://

To clean up, set active false in Maven settings.xml and run these commands.

npm config rm proxy
npm config rm https-proxy
git config --global --unset http.proxy
git config --global --unset https.proxy
git config --global --unset url."http://".insteadOf

Notes: - If you are behind NTLM proxy you can use Cntlm Authentication Proxy. - Replace localhost:3128 with the standard pattern http://user:pwd@host:port.


To package the final distribution including the compressed archive, run:

./mvnw clean package -Pbuild-distr

To build a distribution with specific profiles, run:

./mvnw clean package -Pbuild-distr -Pspark-3.4

The profiles -Pspark-3.4 can be adjusted if you wish to build to a specific spark versions.

The archive is generated under zeppelin-distribution/target directory

Run end-to-end tests

Zeppelin comes with a set of end-to-end acceptance tests driving headless selenium browser

# assumes zeppelin-server running on localhost:8080 (use -Durl=.. to override)
mvn verify

# or take care of starting/stoping zeppelin-server from packaged zeppelin-distribuion/target
mvn verify -P using-packaged-distr