dnet-and/dnet-and-test/src/main/resources/lda_tuning/oozie_app/workflow.xml

<workflow-app name="LDA Tuning WF" xmlns="uri:oozie:workflow:0.5">
    <parameters>
        <property>
            <name>entitiesPath</name>
            <description>the input entity path</description>
        </property>
        <property>
            <name>workingPath</name>
            <description>path for the working directory</description>
        </property>
        <property>
            <name>numPartitions</name>
            <description>number of partitions for the spark files</description>
        </property>
        <property>
            <name>inputFieldJPath</name>
            <description>json path of the input field in the entities</description>
        </property>
        <property>
            <name>vocabularyPath</name>
            <description>location of the vocabulary</description>
        </property>
        <property>
            <name>vocabularyType</name>
            <description>type of the vocabulary: file or tokens</description>
        </property>
        <property>
            <name>trainRatio</name>
            <description>percentage of the data to be used as training set</description>
        </property>
        <property>
            <name>numTopics</name>
            <description>number of topics to which test the LDA model</description>
        </property>
        <property>
            <name>maxIterations</name>
            <description>maximum number of iterations of the LDA algorithm</description>
        </property>
        <property>
            <name>outputModelPath</name>
            <description>location of the best LDA model</description>
        </property>
        <property>
            <name>sparkDriverMemory</name>
            <description>memory for driver process</description>
        </property>
        <property>
            <name>sparkExecutorMemory</name>
            <description>memory for individual executor</description>
        </property>
        <property>
            <name>sparkExecutorCores</name>
            <description>number of cores used by single executor</description>
        </property>
        <property>
            <name>oozieActionShareLibForSpark2</name>
            <description>oozie action sharelib for spark 2.*</description>
        </property>
        <property>
            <name>spark2ExtraListeners</name>
            <value>com.cloudera.spark.lineage.NavigatorAppListener</value>
            <description>spark 2.* extra listeners classname</description>
        </property>
        <property>
            <name>spark2SqlQueryExecutionListeners</name>
            <value>com.cloudera.spark.lineage.NavigatorQueryListener</value>
            <description>spark 2.* sql query execution listeners classname</description>
        </property>
        <property>
            <name>spark2YarnHistoryServerAddress</name>
            <description>spark 2.* yarn history server address</description>
        </property>
        <property>
            <name>spark2EventLogDir</name>
            <description>spark 2.* event log dir location</description>
        </property>
    </parameters>

    <global>
        <job-tracker>${jobTracker}</job-tracker>
        <name-node>${nameNode}</name-node>
        <configuration>
            <property>
                <name>mapreduce.job.queuename</name>
                <value>${queueName}</value>
            </property>
            <property>
                <name>oozie.launcher.mapred.job.queue.name</name>
                <value>${oozieLauncherQueueName}</value>
            </property>
            <property>
                <name>oozie.action.sharelib.for.spark</name>
                <value>${oozieActionShareLibForSpark2}</value>
            </property>
        </configuration>
    </global>

    <start to="resetWorkingPath"/>

    <kill name="Kill">
        <message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
    </kill>

    <action name="resetWorkingPath">
        <fs>
            <delete path="${workingPath}"/>
        </fs>
        <ok to="TokenizeData"/>
        <error to="Kill"/>
    </action>

    <action name="TokenizeData">
        <spark xmlns="uri:oozie:spark-action:0.2">
            <master>yarn</master>
            <mode>cluster</mode>
            <name>Tokenize Data</name>
            <class>eu.dnetlib.jobs.SparkTokenizer</class>
            <jar>dnet-and-test-${projectVersion}.jar</jar>
            <spark-opts>
                --num-executors=32
                --executor-memory=${sparkExecutorMemory}
                --executor-cores=${sparkExecutorCores}
                --driver-memory=${sparkDriverMemory}
                --conf spark.extraListeners=${spark2ExtraListeners}
                --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                --conf spark.sql.shuffle.partitions=3840
                --conf spark.dynamicAllocation.enabled=false
            </spark-opts>
            <arg>--entitiesPath</arg><arg>${entitiesPath}</arg>
            <arg>--inputFieldJPath</arg><arg>${inputFieldJPath}</arg>
            <arg>--workingPath</arg><arg>${workingPath}</arg>
            <arg>--numPartitions</arg><arg>${numPartitions}</arg>
        </spark>
        <ok to="CreateVocabulary"/>
        <error to="Kill"/>
    </action>

    <action name="CreateVocabulary">
        <spark xmlns="uri:oozie:spark-action:0.2">
            <master>yarn</master>
            <mode>cluster</mode>
            <name>Create Vocabulary</name>
            <class>eu.dnetlib.jobs.SparkCreateVocabulary</class>
            <jar>dnet-and-test-${projectVersion}.jar</jar>
            <spark-opts>
                --num-executors=32
                --executor-memory=${sparkExecutorMemory}
                --executor-cores=${sparkExecutorCores}
                --driver-memory=${sparkDriverMemory}
                --conf spark.extraListeners=${spark2ExtraListeners}
                --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                --conf spark.sql.shuffle.partitions=3840
                --conf spark.dynamicAllocation.enabled=true
            </spark-opts>
            <arg>--vocabularyPath</arg><arg>${vocabularyPath}</arg>
            <arg>--vocabularyType</arg><arg>${vocabularyType}</arg>
            <arg>--workingPath</arg><arg>${workingPath}</arg>
            <arg>--numPartitions</arg><arg>${numPartitions}</arg>
        </spark>
        <ok to="CreateCountVectors"/>
        <error to="Kill"/>
    </action>

    <action name="CreateCountVectors">
        <spark xmlns="uri:oozie:spark-action:0.2">
            <master>yarn</master>
            <mode>cluster</mode>
            <name>Create Count Vectors</name>
            <class>eu.dnetlib.jobs.SparkCountVectorizer</class>
            <jar>dnet-and-test-${projectVersion}.jar</jar>
            <spark-opts>
                --executor-memory=${sparkExecutorMemory}
                --executor-cores=${sparkExecutorCores}
                --driver-memory=${sparkDriverMemory}
                --conf spark.extraListeners=${spark2ExtraListeners}
                --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                --conf spark.sql.shuffle.partitions=3840
            </spark-opts>
            <arg>--vocabularyPath</arg><arg>${vocabularyPath}</arg>
            <arg>--workingPath</arg><arg>${workingPath}</arg>
            <arg>--numPartitions</arg><arg>${numPartitions}</arg>
        </spark>
        <ok to="LDATuning"/>
        <error to="Kill"/>
    </action>

    <action name="LDATuning">
        <spark xmlns="uri:oozie:spark-action:0.2">
            <master>yarn</master>
            <mode>cluster</mode>
            <name>LDA Tuning</name>
            <class>eu.dnetlib.jobs.featureextraction.lda.SparkLDATuning</class>
            <jar>dnet-and-test-${projectVersion}.jar</jar>
            <spark-opts>
                --executor-memory=${sparkExecutorMemory}
                --executor-cores=${sparkExecutorCores}
                --driver-memory=${sparkDriverMemory}
                --conf spark.extraListeners=${spark2ExtraListeners}
                --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                --conf spark.sql.shuffle.partitions=3840
            </spark-opts>

            <arg>--trainRatio</arg><arg>${trainRatio}</arg>
            <arg>--numTopics</arg><arg>${numTopics}</arg>
            <arg>--maxIterations</arg><arg>${maxIterations}</arg>
            <arg>--outputModelPath</arg><arg>${outputModelPath}</arg>
            <arg>--workingPath</arg><arg>${workingPath}</arg>
            <arg>--numPartitions</arg><arg>${numPartitions}</arg>
        </spark>
        <ok to="End"/>
        <error to="Kill"/>
    </action>

    <end name="End"/>
</workflow-app>