Dockerization with Dockerfile and running script

Author: hegyibalint

Dockerfile

@@ -0,0 +1,14 @@
+FROM openjdk:8-jdk-alpine
+
+# Download hadoop
+WORKDIR /opt
+RUN apk add bash curl maven python
+RUN curl -L 'http://archive.apache.org/dist/hadoop/core/hadoop-2.6.0/hadoop-2.6.0.tar.gz' | tar -xz
+
+# Copy the project
+COPY . /opt/ldbc_snb_datagen
+WORKDIR /opt/ldbc_snb_datagen
+RUN mvn -DskipTests clean assembly:assembly
+
+ENV HADOOP_CLIENT_OPTS '-Xmx8G'
+CMD /opt/ldbc_snb_datagen/docker_run.sh

docker_run.sh

@@ -0,0 +1,25 @@
+#!/bin/bash
+
+# Variables for the default settings
+DEFAULT_SCALE_FACTOR=snb.interactive.1
+DEFAULT_PERSON_SERIALIZER=ldbc.snb.datagen.serializer.snb.interactive.CSVPersonSerializer
+DEFAULT_INVARIANT_SERIALIZER=ldbc.snb.datagen.serializer.snb.interactive.CSVPersonSerializer
+DEFAULT_PERSON_ACTIVITY_SERIALIZER=ldbc.snb.datagen.serializer.snb.interactive.CSVPersonSerializer
+
+# Parameter serialization
+PARAMS_FILE=params.ini
+echo "ldbc.snb.datagen.generator.scaleFactor:${SCALE_FACTOR:-$DEFAULT_SCALE_FACTOR}" > ${PARAMS_FILE}
+echo "ldbc.snb.datagen.serializer.personSerializer:${PERSON_SERIALIZER:-$DEFAULT_SERIALIZER}" >> ${PARAMS_FILE}
+echo "ldbc.snb.datagen.serializer.invariantSerializer:${INVARIANT_SERIALIZER:-$DEFAULT_INVARIANT_SERIALIZER}" >> ${PARAMS_FILE}
+echo "ldbc.snb.datagen.serializer.personActivitySerializer:${PERSON_ACTIVITY_SERIALIZER:-$DEFAULT_PERSON_ACTIVITY_SERIALIZER}" >> ${PARAMS_FILE}
+
+# Running the generator
+/opt/hadoop-2.6.0/bin/hadoop jar /opt/ldbc_snb_datagen/target/ldbc_snb_datagen-0.2.7-jar-with-dependencies.jar /opt/ldbc_snb_datagen/params.ini
+
+# Cleanup
+rm -f m*personFactors*
+rm -f .m*personFactors*
+rm -f m*activityFactors*
+rm -f .m*activityFactors*
+rm -f m0friendList*
+rm -f .m0friendList*