2020-11-25 14:55:50 +01:00
|
|
|
<workflow-app name="Consistency Graph" xmlns="uri:oozie:workflow:0.5">
|
2020-03-20 13:01:56 +01:00
|
|
|
<parameters>
|
|
|
|
<property>
|
|
|
|
<name>graphBasePath</name>
|
2020-11-25 14:55:32 +01:00
|
|
|
<description>the input graph base path</description>
|
2020-03-20 13:01:56 +01:00
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>workingPath</name>
|
2020-03-23 17:13:30 +01:00
|
|
|
<description>path of the working directory</description>
|
2020-03-20 13:01:56 +01:00
|
|
|
</property>
|
|
|
|
<property>
|
2020-11-25 14:55:32 +01:00
|
|
|
<name>graphOutputPath</name>
|
|
|
|
<description>path of the output graph</description>
|
2020-03-20 13:01:56 +01:00
|
|
|
</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>
|
2020-04-15 21:23:21 +02:00
|
|
|
|
|
|
|
<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>
|
2020-03-20 13:01:56 +01:00
|
|
|
</parameters>
|
|
|
|
|
2020-03-24 17:40:58 +01:00
|
|
|
<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>
|
2020-04-15 21:23:21 +02:00
|
|
|
<property>
|
|
|
|
<name>oozie.action.sharelib.for.spark</name>
|
|
|
|
<value>${oozieActionShareLibForSpark2}</value>
|
|
|
|
</property>
|
2020-03-24 17:40:58 +01:00
|
|
|
</configuration>
|
|
|
|
</global>
|
|
|
|
|
2020-04-17 13:12:44 +02:00
|
|
|
<start to="PropagateRelation"/>
|
2020-03-20 13:01:56 +01:00
|
|
|
|
|
|
|
<kill name="Kill">
|
|
|
|
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
|
|
</kill>
|
|
|
|
|
2020-03-23 11:43:49 +01:00
|
|
|
<action name="PropagateRelation">
|
2020-03-20 13:01:56 +01:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
2020-03-25 11:15:35 +01:00
|
|
|
<master>yarn</master>
|
2020-03-20 13:01:56 +01:00
|
|
|
<mode>cluster</mode>
|
2021-04-21 17:54:51 +02:00
|
|
|
<name>Propagate Relations</name>
|
2020-03-27 10:42:17 +01:00
|
|
|
<class>eu.dnetlib.dhp.oa.dedup.SparkPropagateRelation</class>
|
2020-03-25 15:57:09 +01:00
|
|
|
<jar>dhp-dedup-openaire-${projectVersion}.jar</jar>
|
2020-03-23 11:43:49 +01:00
|
|
|
<spark-opts>
|
2020-04-15 21:23:21 +02:00
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
2020-03-23 11:43:49 +01:00
|
|
|
--driver-memory=${sparkDriverMemory}
|
2020-04-15 21:23:21 +02:00
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.shuffle.partitions=7680
|
2020-03-20 13:01:56 +01:00
|
|
|
</spark-opts>
|
2020-11-25 14:55:32 +01:00
|
|
|
<arg>--graphBasePath</arg><arg>${graphBasePath}</arg>
|
|
|
|
<arg>--o</arg><arg>${graphOutputPath}</arg>
|
|
|
|
<arg>--workingPath</arg><arg>${workingPath}</arg>
|
2020-03-20 13:01:56 +01:00
|
|
|
</spark>
|
2020-11-25 14:55:32 +01:00
|
|
|
<ok to="group_entities"/>
|
2020-04-17 13:12:44 +02:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-11-25 14:55:32 +01:00
|
|
|
<action name="group_entities">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>group graph entities</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.dedup.GroupEntitiesSparkJob</class>
|
|
|
|
<jar>dhp-dedup-openaire-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${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}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.shuffle.partitions=7680
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--graphInputPath</arg><arg>${graphBasePath}</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingPath}/grouped_entities</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="fork_dispatch_entities"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<fork name="fork_dispatch_entities">
|
|
|
|
<path start="dispatch_datasource"/>
|
|
|
|
<path start="dispatch_project"/>
|
|
|
|
<path start="dispatch_organization"/>
|
|
|
|
<path start="dispatch_publication"/>
|
|
|
|
<path start="dispatch_dataset"/>
|
|
|
|
<path start="dispatch_software"/>
|
|
|
|
<path start="dispatch_otherresearchproduct"/>
|
2020-04-17 13:12:44 +02:00
|
|
|
</fork>
|
|
|
|
|
2020-11-25 14:55:32 +01:00
|
|
|
<action name="dispatch_datasource">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Dispatch publications</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.dedup.DispatchEntitiesSparkJob</class>
|
|
|
|
<jar>dhp-dedup-openaire-${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>--inputPath</arg><arg>${workingPath}/grouped_entities</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${graphOutputPath}/datasource</arg>
|
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Datasource</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="wait_dispatch"/>
|
2020-04-17 13:12:44 +02:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-11-25 14:55:32 +01:00
|
|
|
<action name="dispatch_project">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Dispatch project</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.dedup.DispatchEntitiesSparkJob</class>
|
|
|
|
<jar>dhp-dedup-openaire-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${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}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.shuffle.partitions=7680
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--inputPath</arg><arg>${workingPath}/grouped_entities</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${graphOutputPath}/project</arg>
|
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Project</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="wait_dispatch"/>
|
2020-04-17 13:12:44 +02:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-11-25 14:55:32 +01:00
|
|
|
<action name="dispatch_organization">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Dispatch organization</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.dedup.DispatchEntitiesSparkJob</class>
|
|
|
|
<jar>dhp-dedup-openaire-${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>--inputPath</arg><arg>${workingPath}/grouped_entities</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${graphOutputPath}/organization</arg>
|
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Organization</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="wait_dispatch"/>
|
2020-04-17 13:12:44 +02:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-11-25 14:55:32 +01:00
|
|
|
<action name="dispatch_publication">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Dispatch publication</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.dedup.DispatchEntitiesSparkJob</class>
|
|
|
|
<jar>dhp-dedup-openaire-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${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}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.shuffle.partitions=7680
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--inputPath</arg><arg>${workingPath}/grouped_entities</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${graphOutputPath}/publication</arg>
|
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="wait_dispatch"/>
|
2020-04-17 13:12:44 +02:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-11-25 14:55:32 +01:00
|
|
|
<action name="dispatch_dataset">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Dispatch dataset</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.dedup.DispatchEntitiesSparkJob</class>
|
|
|
|
<jar>dhp-dedup-openaire-${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>--inputPath</arg><arg>${workingPath}/grouped_entities</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${graphOutputPath}/dataset</arg>
|
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="wait_dispatch"/>
|
2020-03-20 13:01:56 +01:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-11-25 14:55:32 +01:00
|
|
|
<action name="dispatch_software">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Dispatch software</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.dedup.DispatchEntitiesSparkJob</class>
|
|
|
|
<jar>dhp-dedup-openaire-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${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}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.shuffle.partitions=7680
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--inputPath</arg><arg>${workingPath}/grouped_entities</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${graphOutputPath}/software</arg>
|
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="wait_dispatch"/>
|
2020-04-17 13:12:44 +02:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-11-25 14:55:32 +01:00
|
|
|
<action name="dispatch_otherresearchproduct">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Dispatch otherresearchproduct</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.dedup.DispatchEntitiesSparkJob</class>
|
|
|
|
<jar>dhp-dedup-openaire-${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>--inputPath</arg><arg>${workingPath}/grouped_entities</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${graphOutputPath}/otherresearchproduct</arg>
|
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="wait_dispatch"/>
|
2020-04-17 13:12:44 +02:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-11-25 14:55:32 +01:00
|
|
|
<join name="wait_dispatch" to="End"/>
|
2020-04-17 13:12:44 +02:00
|
|
|
|
2020-03-20 13:01:56 +01:00
|
|
|
<end name="End"/>
|
|
|
|
</workflow-app>
|