dnet-hadoop/dhp-workflows/dhp-graph-provision/src/main/resources/eu/dnetlib/dhp/oa/provision/oozie_app/workflow.xml

<workflow-app name="index_infospace_graph" xmlns="uri:oozie:workflow:0.5">

    <parameters>
        <property>
            <name>inputGraphRootPath</name>
            <description>root location of input materialized graph</description>
        </property>
        <property>
            <name>isLookupUrl</name>
            <description>URL for the isLookup service</description>
        </property>

        <property>
            <name>sparkDriverMemoryForJoining</name>
            <description>memory for driver process</description>
        </property>
        <property>
            <name>sparkExecutorMemoryForJoining</name>
            <description>memory for individual executor</description>
        </property>
        <property>
            <name>sparkExecutorCoresForJoining</name>
            <description>number of cores used by single executor</description>
        </property>
        <property>
            <name>sparkDriverMemoryForIndexing</name>
            <description>memory for driver process</description>
        </property>
        <property>
            <name>sparkExecutorMemoryForIndexing</name>
            <description>memory for individual executor</description>
        </property>
        <property>
            <name>sparkExecutorCoresForIndexing</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>oozie.action.sharelib.for.spark</name>
                <value>${oozieActionShareLibForSpark2}</value>
            </property>
        </configuration>
    </global>

    <start to="reuse_records"/>

    <decision name="reuse_records">
        <switch>
            <case to="prepare_relations">${wf:conf('reuseRecords') eq false}</case>
            <case to="to_solr_index">${wf:conf('reuseRecords') eq true}</case>
            <default to="prepare_relations"/>
        </switch>
    </decision>

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

    <action name="prepare_relations">
        <spark xmlns="uri:oozie:spark-action:0.2">
            <master>yarn</master>
            <mode>cluster</mode>
            <name>PrepareRelations</name>
            <class>eu.dnetlib.dhp.oa.provision.PrepareRelationsJob</class>
            <jar>dhp-graph-provision-${projectVersion}.jar</jar>
            <spark-opts>
                --executor-cores=${sparkExecutorCoresForJoining}
                --executor-memory=${sparkExecutorMemoryForJoining}
                --driver-memory=${sparkDriverMemoryForJoining}
                --conf spark.extraListeners=${spark2ExtraListeners}
                --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                --conf spark.sql.shuffle.partitions=3840
            </spark-opts>
            <arg>--inputRelationsPath</arg><arg>${inputGraphRootPath}/relation</arg>
            <arg>--outputPath</arg><arg>${workingDir}/relation</arg>
            <arg>--relPartitions</arg><arg>3000</arg>
        </spark>
        <ok to="fork_join_related_entities"/>
        <error to="Kill"/>
    </action>

    <fork name="fork_join_related_entities">
        <path start="join_relation_publication"/>
        <path start="join_relation_dataset"/>
        <path start="join_relation_otherresearchproduct"/>
        <path start="join_relation_software"/>
        <path start="join_relation_datasource"/>
        <path start="join_relation_organization"/>
        <path start="join_relation_project"/>
    </fork>

    <action name="join_relation_publication">
        <spark xmlns="uri:oozie:spark-action:0.2">
            <master>yarn</master>
            <mode>cluster</mode>
            <name>Join[relation.target = publication.id]</name>
            <class>eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1</class>
            <jar>dhp-graph-provision-${projectVersion}.jar</jar>
            <spark-opts>
                --executor-cores=${sparkExecutorCoresForJoining}
                --executor-memory=${sparkExecutorMemoryForJoining}
                --driver-memory=${sparkDriverMemoryForJoining}
                --conf spark.extraListeners=${spark2ExtraListeners}
                --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                --conf spark.sql.shuffle.partitions=7680
                --conf spark.network.timeout=${sparkNetworkTimeout}
            </spark-opts>
            <arg>--inputRelationsPath</arg><arg>${workingDir}/relation</arg>
            <arg>--inputEntityPath</arg><arg>${inputGraphRootPath}/publication</arg>
            <arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
            <arg>--outputPath</arg><arg>${workingDir}/join_partial</arg>
        </spark>
        <ok to="wait_joins"/>
        <error to="Kill"/>
    </action>

    <action name="join_relation_dataset">
        <spark xmlns="uri:oozie:spark-action:0.2">
            <master>yarn</master>
            <mode>cluster</mode>
            <name>Join[relation.target = dataset.id]</name>
            <class>eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1</class>
            <jar>dhp-graph-provision-${projectVersion}.jar</jar>
            <spark-opts>
                --executor-cores=${sparkExecutorCoresForJoining}
                --executor-memory=${sparkExecutorMemoryForJoining}
                --driver-memory=${sparkDriverMemoryForJoining}
                --conf spark.extraListeners=${spark2ExtraListeners}
                --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                --conf spark.sql.shuffle.partitions=3840
                --conf spark.network.timeout=${sparkNetworkTimeout}
            </spark-opts>
            <arg>--inputRelationsPath</arg><arg>${workingDir}/relation</arg>
            <arg>--inputEntityPath</arg><arg>${inputGraphRootPath}/dataset</arg>
            <arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
            <arg>--outputPath</arg><arg>${workingDir}/join_partial</arg>
        </spark>
        <ok to="wait_joins"/>
        <error to="Kill"/>
    </action>

    <action name="join_relation_otherresearchproduct">
        <spark xmlns="uri:oozie:spark-action:0.2">
            <master>yarn</master>
            <mode>cluster</mode>
            <name>Join[relation.target = otherresearchproduct.id]</name>
            <class>eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1</class>
            <jar>dhp-graph-provision-${projectVersion}.jar</jar>
            <spark-opts>
                --executor-cores=${sparkExecutorCoresForJoining}
                --executor-memory=${sparkExecutorMemoryForJoining}
                --driver-memory=${sparkDriverMemoryForJoining}
                --conf spark.extraListeners=${spark2ExtraListeners}
                --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                --conf spark.sql.shuffle.partitions=3840
                --conf spark.network.timeout=${sparkNetworkTimeout}
            </spark-opts>
            <arg>--inputRelationsPath</arg><arg>${workingDir}/relation</arg>
            <arg>--inputEntityPath</arg><arg>${inputGraphRootPath}/otherresearchproduct</arg>
            <arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
            <arg>--outputPath</arg><arg>${workingDir}/join_partial</arg>
        </spark>
        <ok to="wait_joins"/>
        <error to="Kill"/>
    </action>

    <action name="join_relation_software">
        <spark xmlns="uri:oozie:spark-action:0.2">
            <master>yarn</master>
            <mode>cluster</mode>
            <name>Join[relation.target = software.id]</name>
            <class>eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1</class>
            <jar>dhp-graph-provision-${projectVersion}.jar</jar>
            <spark-opts>
                --executor-cores=${sparkExecutorCoresForJoining}
                --executor-memory=${sparkExecutorMemoryForJoining}
                --driver-memory=${sparkDriverMemoryForJoining}
                --conf spark.extraListeners=${spark2ExtraListeners}
                --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                --conf spark.sql.shuffle.partitions=3840
                --conf spark.network.timeout=${sparkNetworkTimeout}
            </spark-opts>
            <arg>--inputRelationsPath</arg><arg>${workingDir}/relation</arg>
            <arg>--inputEntityPath</arg><arg>${inputGraphRootPath}/software</arg>
            <arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
            <arg>--outputPath</arg><arg>${workingDir}/join_partial</arg>
        </spark>
        <ok to="wait_joins"/>
        <error to="Kill"/>
    </action>

    <action name="join_relation_datasource">
        <spark xmlns="uri:oozie:spark-action:0.2">
            <master>yarn</master>
            <mode>cluster</mode>
            <name>Join[relation.target = datasource.id]</name>
            <class>eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1</class>
            <jar>dhp-graph-provision-${projectVersion}.jar</jar>
            <spark-opts>
                --executor-cores=${sparkExecutorCoresForJoining}
                --executor-memory=${sparkExecutorMemoryForJoining}
                --driver-memory=${sparkDriverMemoryForJoining}
                --conf spark.extraListeners=${spark2ExtraListeners}
                --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                --conf spark.sql.shuffle.partitions=3840
                --conf spark.network.timeout=${sparkNetworkTimeout}
            </spark-opts>
            <arg>--inputRelationsPath</arg><arg>${workingDir}/relation</arg>
            <arg>--inputEntityPath</arg><arg>${inputGraphRootPath}/datasource</arg>
            <arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Datasource</arg>
            <arg>--outputPath</arg><arg>${workingDir}/join_partial</arg>
        </spark>
        <ok to="wait_joins"/>
        <error to="Kill"/>
    </action>

    <action name="join_relation_organization">
        <spark xmlns="uri:oozie:spark-action:0.2">
            <master>yarn</master>
            <mode>cluster</mode>
            <name>Join[relation.target = organization.id]</name>
            <class>eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1</class>
            <jar>dhp-graph-provision-${projectVersion}.jar</jar>
            <spark-opts>
                --executor-cores=${sparkExecutorCoresForJoining}
                --executor-memory=${sparkExecutorMemoryForJoining}
                --driver-memory=${sparkDriverMemoryForJoining}
                --conf spark.extraListeners=${spark2ExtraListeners}
                --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                --conf spark.sql.shuffle.partitions=3840
                --conf spark.network.timeout=${sparkNetworkTimeout}
            </spark-opts>
            <arg>--inputRelationsPath</arg><arg>${workingDir}/relation</arg>
            <arg>--inputEntityPath</arg><arg>${inputGraphRootPath}/organization</arg>
            <arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Organization</arg>
            <arg>--outputPath</arg><arg>${workingDir}/join_partial</arg>
        </spark>
        <ok to="wait_joins"/>
        <error to="Kill"/>
    </action>

    <action name="join_relation_project">
        <spark xmlns="uri:oozie:spark-action:0.2">
            <master>yarn</master>
            <mode>cluster</mode>
            <name>Join[relation.target = project.id]</name>
            <class>eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1</class>
            <jar>dhp-graph-provision-${projectVersion}.jar</jar>
            <spark-opts>
                --executor-cores=${sparkExecutorCoresForJoining}
                --executor-memory=${sparkExecutorMemoryForJoining}
                --driver-memory=${sparkDriverMemoryForJoining}
                --conf spark.extraListeners=${spark2ExtraListeners}
                --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                --conf spark.sql.shuffle.partitions=3840
                --conf spark.network.timeout=${sparkNetworkTimeout}
            </spark-opts>
            <arg>--inputRelationsPath</arg><arg>${workingDir}/relation</arg>
            <arg>--inputEntityPath</arg><arg>${inputGraphRootPath}/project</arg>
            <arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Project</arg>
            <arg>--outputPath</arg><arg>${workingDir}/join_partial</arg>
        </spark>
        <ok to="wait_joins"/>
        <error to="Kill"/>
    </action>

    <join name="wait_joins" to="join_all_entities"/>


    <action name="join_all_entities">
        <spark xmlns="uri:oozie:spark-action:0.2">
            <master>yarn</master>
            <mode>cluster</mode>
            <name>Join[entities.id = relatedEntity.source]</name>
            <class>eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase2</class>
            <jar>dhp-graph-provision-${projectVersion}.jar</jar>
            <spark-opts>
                --executor-cores=${sparkExecutorCoresForJoining}
                --executor-memory=${sparkExecutorMemoryForJoining}
                --driver-memory=${sparkDriverMemoryForJoining}
                --conf spark.extraListeners=${spark2ExtraListeners}
                --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                --conf spark.sql.shuffle.partitions=7680
                --conf spark.network.timeout=${sparkNetworkTimeout}
            </spark-opts>
            <arg>--inputGraphRootPath</arg><arg>${inputGraphRootPath}</arg>
            <arg>--inputRelatedEntitiesPath</arg><arg>${workingDir}/join_partial</arg>
            <arg>--outputPath</arg><arg>${workingDir}/join_entities</arg>
            <arg>--numPartitions</arg><arg>12000</arg>
        </spark>
        <ok to="adjancency_lists"/>
        <error to="Kill"/>
    </action>

    <action name="adjancency_lists">
        <spark xmlns="uri:oozie:spark-action:0.2">
            <master>yarn</master>
            <mode>cluster</mode>
            <name>build_adjacency_lists</name>
            <class>eu.dnetlib.dhp.oa.provision.AdjacencyListBuilderJob</class>
            <jar>dhp-graph-provision-${projectVersion}.jar</jar>
            <spark-opts>
                --executor-cores=${sparkExecutorCoresForJoining}
                --executor-memory=${sparkExecutorMemoryForJoining}
                --driver-memory=${sparkDriverMemoryForJoining}
                --conf spark.extraListeners=${spark2ExtraListeners}
                --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                --conf spark.sql.shuffle.partitions=7680
                --conf spark.network.timeout=${sparkNetworkTimeout}
            </spark-opts>
            <arg>--inputPath</arg> <arg>${workingDir}/join_entities</arg>
            <arg>--outputPath</arg><arg>${workingDir}/joined</arg>
        </spark>
        <ok to="convert_to_xml"/>
        <error to="Kill"/>
    </action>

    <action name="convert_to_xml">
        <spark xmlns="uri:oozie:spark-action:0.2">
            <master>yarn</master>
            <mode>cluster</mode>
            <name>convert_to_xml</name>
            <class>eu.dnetlib.dhp.oa.provision.XmlConverterJob</class>
            <jar>dhp-graph-provision-${projectVersion}.jar</jar>
            <spark-opts>
                --executor-cores=${sparkExecutorCoresForJoining}
                --executor-memory=${sparkExecutorMemoryForJoining}
                --driver-memory=${sparkDriverMemoryForJoining}
                --conf spark.extraListeners=${spark2ExtraListeners}
                --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                --conf spark.sql.shuffle.partitions=3840
                --conf spark.network.timeout=${sparkNetworkTimeout}
            </spark-opts>
            <arg>--inputPath</arg><arg>${workingDir}/joined</arg>
            <arg>--outputPath</arg><arg>${workingDir}/xml</arg>
            <arg>--isLookupUrl</arg><arg>${isLookupUrl}</arg>
            <arg>--otherDsTypeId</arg><arg>${otherDsTypeId}</arg>
        </spark>
        <ok to="to_solr_index"/>
        <error to="Kill"/>
    </action>

    <action name="to_solr_index">
        <spark xmlns="uri:oozie:spark-action:0.2">
            <master>yarn</master>
            <mode>cluster</mode>
            <name>to_solr_index</name>
            <class>eu.dnetlib.dhp.oa.provision.XmlIndexingJob</class>
            <jar>dhp-graph-provision-${projectVersion}.jar</jar>
            <spark-opts>
                --executor-memory=${sparkExecutorMemoryForIndexing}
                --driver-memory=${sparkDriverMemoryForIndexing}
                --conf spark.dynamicAllocation.enabled=true
                --conf spark.dynamicAllocation.maxExecutors=${sparkExecutorCoresForIndexing}
                --conf spark.extraListeners=${spark2ExtraListeners}
                --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                --conf spark.speculation=false
                --conf spark.hadoop.mapreduce.map.speculative=false
                --conf spark.hadoop.mapreduce.reduce.speculative=false
            </spark-opts>
            <arg>--inputPath</arg><arg>${workingDir}/xml</arg>
            <arg>--isLookupUrl</arg> <arg>${isLookupUrl}</arg>
            <arg>--format</arg><arg>${format}</arg>
            <arg>--batchSize</arg><arg>${batchSize}</arg>
        </spark>
        <ok to="End"/>
        <error to="Kill"/>
    </action>

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