2020-11-13 10:05:12 +01:00
|
|
|
<workflow-app name="group graph entities and relations" xmlns="uri:oozie:workflow:0.5">
|
2020-11-13 10:05:02 +01:00
|
|
|
|
|
|
|
<parameters>
|
|
|
|
<property>
|
|
|
|
<name>graphInputPath</name>
|
|
|
|
<description>the graph root input path</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>outputPath</name>
|
|
|
|
<description>the graph root output path</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>
|
|
|
|
|
2020-11-13 10:05:12 +01:00
|
|
|
<start to="group_entities"/>
|
2020-11-13 10:05:02 +01:00
|
|
|
|
|
|
|
<kill name="Kill">
|
|
|
|
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
|
|
</kill>
|
|
|
|
|
2020-11-13 10:05:12 +01:00
|
|
|
<action name="group_entities">
|
2020-11-13 10:05:02 +01:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
2020-11-13 10:05:12 +01:00
|
|
|
<name>group graph entities and relations</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.graph.groupbyid.GroupEntitiesAndRelationsSparkJob</class>
|
2020-11-13 10:05:02 +01:00
|
|
|
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
--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=7680
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--graphInputPath</arg><arg>${graphInputPath}</arg>
|
2020-11-13 10:05:12 +01:00
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/grouped_entities</arg>
|
2020-11-13 10:05:02 +01:00
|
|
|
</spark>
|
2020-11-13 10:05:12 +01:00
|
|
|
<ok to="fork_dispatch_entities"/>
|
2020-11-13 10:05:02 +01:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-11-13 10:05:12 +01:00
|
|
|
<fork name="fork_dispatch_entities">
|
|
|
|
<path start="dispatch_publication"/>
|
|
|
|
<path start="dispatch_dataset"/>
|
|
|
|
<path start="dispatch_software"/>
|
|
|
|
<path start="dispatch_otherresearchproduct"/>
|
|
|
|
<path start="dispatch_datasource"/>
|
|
|
|
<path start="dispatch_organization"/>
|
|
|
|
<path start="dispatch_project"/>
|
|
|
|
<path start="dispatch_relation"/>
|
|
|
|
</fork>
|
|
|
|
|
|
|
|
<action name="dispatch_publication">
|
2020-11-13 10:05:02 +01:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
2020-11-13 10:05:12 +01:00
|
|
|
<name>Dispatch publications</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.graph.groupbyid.DispatchEntitiesSparkJob</class>
|
2020-11-13 10:05:02 +01:00
|
|
|
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
--driver-memory=${sparkDriverMemory}
|
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.shuffle.partitions=7680
|
|
|
|
</spark-opts>
|
2020-11-13 10:05:12 +01:00
|
|
|
<arg>--entitiesPath</arg><arg>${workingDir}/grouped_entities</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${outputPath}/publication</arg>
|
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
|
2020-11-13 10:05:02 +01:00
|
|
|
</spark>
|
2020-11-13 10:05:12 +01:00
|
|
|
<ok to="wait_dispatch"/>
|
2020-11-13 10:05:02 +01:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-11-13 10:05:12 +01:00
|
|
|
<action name="dispatch_dataset">
|
2020-11-13 10:05:02 +01:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
2020-11-13 10:05:12 +01:00
|
|
|
<name>Dispatch datasets</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.graph.groupbyid.DispatchEntitiesSparkJob</class>
|
2020-11-13 10:05:02 +01:00
|
|
|
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
--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=7680
|
|
|
|
</spark-opts>
|
2020-11-13 10:05:12 +01:00
|
|
|
<arg>--entitiesPath</arg><arg>${workingDir}/grouped_entities</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${outputPath}/dataset</arg>
|
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
|
2020-11-13 10:05:02 +01:00
|
|
|
</spark>
|
2020-11-13 10:05:12 +01:00
|
|
|
<ok to="wait_dispatch"/>
|
2020-11-13 10:05:02 +01:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-11-13 10:05:12 +01:00
|
|
|
<action name="dispatch_software">
|
2020-11-13 10:05:02 +01:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
2020-11-13 10:05:12 +01:00
|
|
|
<name>Dispatch softwares</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.graph.groupbyid.DispatchEntitiesSparkJob</class>
|
2020-11-13 10:05:02 +01:00
|
|
|
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
--driver-memory=${sparkDriverMemory}
|
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.shuffle.partitions=7680
|
|
|
|
</spark-opts>
|
2020-11-13 10:05:12 +01:00
|
|
|
<arg>--entitiesPath</arg><arg>${workingDir}/grouped_entities</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${outputPath}/software</arg>
|
2020-11-13 10:05:02 +01:00
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
|
|
|
|
</spark>
|
2020-11-13 10:05:12 +01:00
|
|
|
<ok to="wait_dispatch"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="dispatch_otherresearchproduct">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Dispatch otherresearchproducts</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.graph.groupbyid.DispatchEntitiesSparkJob</class>
|
|
|
|
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
--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=7680
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--entitiesPath</arg><arg>${workingDir}/grouped_entities</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${outputPath}/otherresearchproduct</arg>
|
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="wait_dispatch"/>
|
2020-11-13 10:05:02 +01:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-11-13 10:05:12 +01:00
|
|
|
<action name="dispatch_datasource">
|
2020-11-13 10:05:02 +01:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
2020-11-13 10:05:12 +01:00
|
|
|
<name>Dispatch datasources</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.graph.groupbyid.DispatchEntitiesSparkJob</class>
|
2020-11-13 10:05:02 +01:00
|
|
|
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
--driver-memory=${sparkDriverMemory}
|
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.shuffle.partitions=7680
|
|
|
|
</spark-opts>
|
2020-11-13 10:05:12 +01:00
|
|
|
<arg>--entitiesPath</arg><arg>${workingDir}/grouped_entities</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${outputPath}/datasource</arg>
|
2020-11-13 10:05:02 +01:00
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Datasource</arg>
|
|
|
|
</spark>
|
2020-11-13 10:05:12 +01:00
|
|
|
<ok to="wait_dispatch"/>
|
2020-11-13 10:05:02 +01:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-11-13 10:05:12 +01:00
|
|
|
<action name="dispatch_organization">
|
2020-11-13 10:05:02 +01:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
2020-11-13 10:05:12 +01:00
|
|
|
<name>Dispatch organizations</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.graph.groupbyid.DispatchEntitiesSparkJob</class>
|
2020-11-13 10:05:02 +01:00
|
|
|
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
--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=7680
|
|
|
|
</spark-opts>
|
2020-11-13 10:05:12 +01:00
|
|
|
<arg>--entitiesPath</arg><arg>${workingDir}/grouped_entities</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${outputPath}/organization</arg>
|
2020-11-13 10:05:02 +01:00
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Organization</arg>
|
|
|
|
</spark>
|
2020-11-13 10:05:12 +01:00
|
|
|
<ok to="wait_dispatch"/>
|
2020-11-13 10:05:02 +01:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-11-13 10:05:12 +01:00
|
|
|
<action name="dispatch_project">
|
2020-11-13 10:05:02 +01:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
2020-11-13 10:05:12 +01:00
|
|
|
<name>Dispatch project</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.graph.groupbyid.DispatchEntitiesSparkJob</class>
|
2020-11-13 10:05:02 +01:00
|
|
|
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
--driver-memory=${sparkDriverMemory}
|
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.shuffle.partitions=7680
|
|
|
|
</spark-opts>
|
2020-11-13 10:05:12 +01:00
|
|
|
<arg>--entitiesPath</arg><arg>${workingDir}/grouped_entities</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${outputPath}/project</arg>
|
2020-11-13 10:05:02 +01:00
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Project</arg>
|
|
|
|
</spark>
|
2020-11-13 10:05:12 +01:00
|
|
|
<ok to="wait_dispatch"/>
|
2020-11-13 10:05:02 +01:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-11-13 10:05:12 +01:00
|
|
|
<action name="dispatch_relation">
|
2020-11-13 10:05:02 +01:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
2020-11-13 10:05:12 +01:00
|
|
|
<name>Dispatch relations</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.graph.groupbyid.DispatchEntitiesSparkJob</class>
|
2020-11-13 10:05:02 +01:00
|
|
|
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
--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=7680
|
|
|
|
</spark-opts>
|
2020-11-13 10:05:12 +01:00
|
|
|
<arg>--entitiesPath</arg><arg>${workingDir}/grouped_entities</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${outputPath}/relation</arg>
|
2020-11-13 10:05:02 +01:00
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Relation</arg>
|
|
|
|
</spark>
|
2020-11-13 10:05:12 +01:00
|
|
|
<ok to="wait_dispatch"/>
|
2020-11-13 10:05:02 +01:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-11-13 10:05:12 +01:00
|
|
|
<join name="wait_dispatch" to="End"/>
|
2020-11-13 10:05:02 +01:00
|
|
|
|
|
|
|
<end name="End"/>
|
|
|
|
</workflow-app>
|