How can I switch the Spark version in Zeppelin to use Spark 3.x

Question

I'm trying to set up a Dockerised version of Spark and Zeppelin but I cannot seem to understand how to switch the Zeppelin version to the 3.x version of Spark.

I'm using the default Zeppelin image from Docker Hub. Here's an excerpt from my docker-compose.yml.

  zeppelin:
    image: apache/zeppelin:0.9.0
    container_name: zeppelin
    #depends_on:
    #  - spark-master
    ports:
      - "8083:8080"

If I access Zeppelin (at localhost:8083), and execute spark.version, it still reads the version as 2.4.5.

How do I change the spark version in Zeppelin? I can see a fair number of versions supported but the docs don't clarify how to switch versions.https://github.com/apache/zeppelin/blob/master/spark/spark-shims/src/main/scala/org/apache/zeppelin/spark/SparkVersion.java#L25

user_6551426 · Answer

You can run spark in a separate container and point the spark master to it, Another easy way is to build your image with a Spark on top of Zeppelin

Create a Dockerfile file with zeppelin as the base image

FROM apache/zeppelin:0.9.0
ENV SPARK_VERSION=3.0.0
ENV HADOOP_VERSION=3.2
ENV SPARK_INSTALL_ROOT=/spark

ENV SPARK_HOME=${SPARK_INSTALL_ROOT}/spark-${SPARK_VERSION}-bin-hadoop${HADOOP_VERSION}

USER root
RUN mkdir "${SPARK_INSTALL_ROOT}"
USER $USER

RUN cd "${SPARK_INSTALL_ROOT}" && \
    wget --show-progress https://archive.apache.org/dist/spark/spark-$SPARK_VERSION/spark-$SPARK_VERSION-bin-hadoop${HADOOP_VERSION}.tgz && \
    tar -xzf spark-${SPARK_VERSION}-bin-hadoop${HADOOP_VERSION}.tgz

Now Use this image to run with docker-compose, You can create the image with a tag and use it here or you can directly refer to Dockerfile as below version: '2'

services:
  zeppelin:
    image: zeppelin-spark
    build:
      context: .
      dockerfile: Dockerfile
    container_name: zeppelin
    ports:
      - "8083:8080"

Now run docker-compose up -d make sure both files are in the same directory or feel free to adjust the path in the context

Then you see the version as 3.0.0

user_1888799 · Answer

You need to set SPARK_HOME & spark_master with a proper one. Please follow instructions in the Zeppelin's documentation:

https://zeppelin.apache.org/docs/latest/interpreter/spark.html

How can I switch the Spark version in Zeppelin to use Spark 3.x

2 Answers