Dockerfile.spark

FROM debian:stretch

RUN apt-get update && apt-get install -yq --no-install-recommends --force-yes \
    curl \
    git \
    openjdk-8-jdk \
    maven \
    python2.7 python2.7-setuptools \
    python3 python3-setuptools \
    r-base \
    r-base-core && \
    rm -rf /var/lib/apt/lists/*

RUN easy_install3 pip py4j
RUN pip install --upgrade setuptools

ENV PYTHONHASHSEED 0
ENV PYTHONIOENCODING UTF-8
ENV PIP_DISABLE_PIP_VERSION_CHECK 1

ENV SPARK_BUILD_VERSION 2.3.3
ENV SPARK_HOME /apps/spark-$SPARK_BUILD_VERSION
ENV SPARK_BUILD_PATH /apps/build/spark
RUN mkdir -p /apps/build && \
  cd /apps/build && \
  git clone https://github.com/apache/spark.git spark && \
  cd $SPARK_BUILD_PATH && \
  git checkout v$SPARK_BUILD_VERSION && \
  dev/make-distribution.sh --name spark-$SPARK_BUILD_VERSION -Phive -Phive-thriftserver -Pyarn && \
  cp -r /apps/build/spark/dist $SPARK_HOME && \
  rm -rf $SPARK_BUILD_PATH

ENV LIVY_BUILD_VERSION 0.6.0-incubating
ENV LIVY_APP_PATH /apps/apache-livy-$LIVY_BUILD_VERSION-bin
ENV LIVY_BUILD_PATH /apps/build/livy
RUN cd /apps/build && \
	git clone https://github.com/apache/incubator-livy.git livy && \
	cd $LIVY_BUILD_PATH && \
  git checkout v$LIVY_BUILD_VERSION-rc2 && \
    mvn -DskipTests -Dspark.version=$SPARK_BUILD_VERSION clean package && \
    ls -al $LIVY_BUILD_PATH && ls -al $LIVY_BUILD_PATH/assembly && ls -al $LIVY_BUILD_PATH/assembly/target && \
    unzip $LIVY_BUILD_PATH/assembly/target/apache-livy-${LIVY_BUILD_VERSION}-bin.zip -d /apps && \
    rm -rf $LIVY_BUILD_PATH && \
	mkdir -p $LIVY_APP_PATH/upload && \
  mkdir -p $LIVY_APP_PATH/logs

RUN pip install matplotlib
RUN pip install pandas

EXPOSE 8998

CMD $LIVY_APP_PATH/bin/livy-server