2020-02-28 18:26:04 +01:00
|
|
|
<workflow-app name="result_to_community_from_semrel_propagation" xmlns="uri:oozie:workflow:0.5">
|
2020-05-08 12:33:31 +02:00
|
|
|
<parameters>
|
|
|
|
<property>
|
|
|
|
<name>sourcePath</name>
|
|
|
|
<description>the source path</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>allowedsemrels</name>
|
|
|
|
<description>the semantic relationships allowed for propagation</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>isLookUpUrl</name>
|
|
|
|
<description>the isLookup service endpoint</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>outputPath</name>
|
|
|
|
<description>the output path</description>
|
|
|
|
</property>
|
|
|
|
</parameters>
|
2020-02-28 18:26:04 +01:00
|
|
|
|
2020-05-08 12:33:31 +02:00
|
|
|
<start to="reset_outputpath"/>
|
2020-04-27 10:48:31 +02:00
|
|
|
|
|
|
|
<kill name="Kill">
|
|
|
|
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
|
|
</kill>
|
2020-05-08 12:33:31 +02:00
|
|
|
|
|
|
|
<action name="reset_outputpath">
|
2020-04-27 10:48:31 +02:00
|
|
|
<fs>
|
2020-05-08 12:33:31 +02:00
|
|
|
<delete path="${outputPath}"/>
|
|
|
|
<mkdir path="${outputPath}"/>
|
2020-04-27 10:48:31 +02:00
|
|
|
</fs>
|
2020-04-30 12:45:28 +02:00
|
|
|
<ok to="copy_entities"/>
|
2020-04-27 10:48:31 +02:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
2020-05-08 12:33:31 +02:00
|
|
|
|
2020-04-30 12:45:28 +02:00
|
|
|
<fork name="copy_entities">
|
|
|
|
<path start="copy_relation"/>
|
|
|
|
<path start="copy_organization"/>
|
|
|
|
<path start="copy_projects"/>
|
|
|
|
<path start="copy_datasources"/>
|
|
|
|
</fork>
|
|
|
|
|
2020-04-27 10:48:31 +02:00
|
|
|
<action name="copy_relation">
|
|
|
|
<distcp xmlns="uri:oozie:distcp-action:0.2">
|
|
|
|
<job-tracker>${jobTracker}</job-tracker>
|
|
|
|
<name-node>${nameNode}</name-node>
|
|
|
|
<arg>${nameNode}/${sourcePath}/relation</arg>
|
2020-04-30 12:45:28 +02:00
|
|
|
<arg>${nameNode}/${outputPath}/relation</arg>
|
|
|
|
</distcp>
|
|
|
|
<ok to="copy_wait"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="copy_organization">
|
|
|
|
<distcp xmlns="uri:oozie:distcp-action:0.2">
|
|
|
|
<job-tracker>${jobTracker}</job-tracker>
|
|
|
|
<name-node>${nameNode}</name-node>
|
|
|
|
<arg>${nameNode}/${sourcePath}/organization</arg>
|
|
|
|
<arg>${nameNode}/${outputPath}/organization</arg>
|
2020-04-27 10:48:31 +02:00
|
|
|
</distcp>
|
2020-04-30 12:45:28 +02:00
|
|
|
<ok to="copy_wait"/>
|
2020-04-27 10:48:31 +02:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
2020-02-28 18:26:04 +01:00
|
|
|
|
2020-04-30 12:45:28 +02:00
|
|
|
<action name="copy_projects">
|
|
|
|
<distcp xmlns="uri:oozie:distcp-action:0.2">
|
|
|
|
<job-tracker>${jobTracker}</job-tracker>
|
|
|
|
<name-node>${nameNode}</name-node>
|
|
|
|
<arg>${nameNode}/${sourcePath}/project</arg>
|
|
|
|
<arg>${nameNode}/${outputPath}/project</arg>
|
|
|
|
</distcp>
|
|
|
|
<ok to="copy_wait"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="copy_datasources">
|
|
|
|
<distcp xmlns="uri:oozie:distcp-action:0.2">
|
|
|
|
<job-tracker>${jobTracker}</job-tracker>
|
|
|
|
<name-node>${nameNode}</name-node>
|
|
|
|
<arg>${nameNode}/${sourcePath}/datasource</arg>
|
|
|
|
<arg>${nameNode}/${outputPath}/datasource</arg>
|
|
|
|
</distcp>
|
|
|
|
<ok to="copy_wait"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<join name="copy_wait" to="fork_prepare_assoc_step1"/>
|
|
|
|
|
2020-04-27 10:48:31 +02:00
|
|
|
<fork name="fork_prepare_assoc_step1">
|
|
|
|
<path start="join_prepare_publication"/>
|
|
|
|
<path start="join_prepare_dataset"/>
|
|
|
|
<path start="join_prepare_otherresearchproduct"/>
|
|
|
|
<path start="join_prepare_software"/>
|
|
|
|
</fork>
|
2020-02-28 18:26:04 +01:00
|
|
|
|
2020-04-27 10:48:31 +02:00
|
|
|
<action name="join_prepare_publication">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>ResultToCommunitySemRel-PreparePhase1-Publications</name>
|
|
|
|
<class>eu.dnetlib.dhp.resulttocommunityfromsemrel.PrepareResultCommunitySetStep1</class>
|
2020-05-12 15:49:28 +02:00
|
|
|
<jar>dhp-enrichment-${projectVersion}.jar</jar>
|
2020-04-27 10:48:31 +02:00
|
|
|
<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.dynamicAllocation.enabled=true
|
|
|
|
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}</arg>
|
|
|
|
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
|
|
|
|
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/preparedInfo/targetCommunityAssoc</arg>
|
|
|
|
<arg>--allowedsemrels</arg><arg>${allowedsemrels}</arg>
|
2020-05-05 10:03:32 +02:00
|
|
|
<arg>--isLookUpUrl</arg><arg>${isLookUpUrl}</arg>
|
2020-04-27 10:48:31 +02:00
|
|
|
</spark>
|
|
|
|
<ok to="wait"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
2020-05-08 12:33:31 +02:00
|
|
|
|
2020-04-27 10:48:31 +02:00
|
|
|
<action name="join_prepare_dataset">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>ResultToCommunitySemRel-PreparePhase1-Dataset</name>
|
|
|
|
<class>eu.dnetlib.dhp.resulttocommunityfromsemrel.PrepareResultCommunitySetStep1</class>
|
2020-05-12 15:49:28 +02:00
|
|
|
<jar>dhp-enrichment-${projectVersion}.jar</jar>
|
2020-04-27 10:48:31 +02:00
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${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}${spark2EventLogDir}
|
|
|
|
--conf spark.dynamicAllocation.enabled=true
|
|
|
|
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}</arg>
|
|
|
|
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
|
|
|
|
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/preparedInfo/targetCommunityAssoc</arg>
|
|
|
|
<arg>--allowedsemrels</arg><arg>${allowedsemrels}</arg>
|
2020-05-05 10:03:32 +02:00
|
|
|
<arg>--isLookUpUrl</arg><arg>${isLookUpUrl}</arg>
|
2020-04-27 10:48:31 +02:00
|
|
|
</spark>
|
|
|
|
<ok to="wait"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
2020-05-12 15:49:28 +02:00
|
|
|
|
2020-04-27 10:48:31 +02:00
|
|
|
<action name="join_prepare_otherresearchproduct">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>ResultToCommunitySemRel-PreparePhase1-ORP</name>
|
|
|
|
<class>eu.dnetlib.dhp.resulttocommunityfromsemrel.PrepareResultCommunitySetStep1</class>
|
2020-05-12 15:49:28 +02:00
|
|
|
<jar>dhp-enrichment-${projectVersion}.jar</jar>
|
2020-04-27 10:48:31 +02:00
|
|
|
<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.dynamicAllocation.enabled=true
|
|
|
|
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}</arg>
|
|
|
|
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
|
|
|
|
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/preparedInfo/targetCommunityAssoc</arg>
|
|
|
|
<arg>--allowedsemrels</arg><arg>${allowedsemrels}</arg>
|
2020-05-05 10:03:32 +02:00
|
|
|
<arg>--isLookUpUrl</arg><arg>${isLookUpUrl}</arg>
|
2020-04-27 10:48:31 +02:00
|
|
|
</spark>
|
|
|
|
<ok to="wait"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
2020-05-08 12:33:31 +02:00
|
|
|
|
2020-04-27 10:48:31 +02:00
|
|
|
<action name="join_prepare_software">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>ResultToCommunitySemRel-PreparePhase1-Software</name>
|
|
|
|
<class>eu.dnetlib.dhp.resulttocommunityfromsemrel.PrepareResultCommunitySetStep1</class>
|
2020-05-12 15:49:28 +02:00
|
|
|
<jar>dhp-enrichment-${projectVersion}.jar</jar>
|
2020-04-27 10:48:31 +02:00
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${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}${spark2EventLogDir}
|
|
|
|
--conf spark.dynamicAllocation.enabled=true
|
|
|
|
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}</arg>
|
|
|
|
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
|
|
|
|
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/preparedInfo/targetCommunityAssoc</arg>
|
|
|
|
<arg>--allowedsemrels</arg><arg>${allowedsemrels}</arg>
|
2020-05-05 10:03:32 +02:00
|
|
|
<arg>--isLookUpUrl</arg><arg>${isLookUpUrl}</arg>
|
2020-04-27 10:48:31 +02:00
|
|
|
</spark>
|
|
|
|
<ok to="wait"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<join name="wait" to="prepare_assoc_step2"/>
|
|
|
|
|
|
|
|
<action name="prepare_assoc_step2">
|
2020-05-08 12:33:31 +02:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>ResultToCommunityEmRelPropagation-PreparePhase2</name>
|
|
|
|
<class>eu.dnetlib.dhp.resulttocommunityfromsemrel.PrepareResultCommunitySetStep2</class>
|
2020-05-12 15:49:28 +02:00
|
|
|
<jar>dhp-enrichment-${projectVersion}.jar</jar>
|
2020-05-08 12:33:31 +02:00
|
|
|
<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.dynamicAllocation.enabled=true
|
|
|
|
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--sourcePath</arg><arg>${workingDir}/preparedInfo/targetCommunityAssoc</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/preparedInfo/mergedCommunityAssoc</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="fork-join-exec-propagation"/>
|
|
|
|
<error to="Kill"/>
|
2020-04-27 10:48:31 +02:00
|
|
|
</action>
|
|
|
|
|
|
|
|
<fork name="fork-join-exec-propagation">
|
|
|
|
<path start="join_propagate_publication"/>
|
|
|
|
<path start="join_propagate_dataset"/>
|
|
|
|
<path start="join_propagate_otherresearchproduct"/>
|
|
|
|
<path start="join_propagate_software"/>
|
|
|
|
</fork>
|
|
|
|
|
|
|
|
<action name="join_propagate_publication">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Result2CommunitySemRelPropagation-Publication</name>
|
2020-05-07 18:22:26 +02:00
|
|
|
<class>eu.dnetlib.dhp.resulttocommunityfromsemrel.SparkResultToCommunityThroughSemRelJob</class>
|
2020-05-12 15:49:28 +02:00
|
|
|
<jar>dhp-enrichment-${projectVersion}.jar</jar>
|
2020-04-27 10:48:31 +02:00
|
|
|
<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.dynamicAllocation.enabled=true
|
|
|
|
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--preparedInfoPath</arg><arg>${workingDir}/preparedInfo/mergedCommunityAssoc</arg>
|
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}/publication</arg>
|
|
|
|
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
|
|
|
|
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
|
2020-04-30 12:45:28 +02:00
|
|
|
<arg>--outputPath</arg><arg>${outputPath}/publication</arg>
|
2020-05-05 10:03:32 +02:00
|
|
|
<arg>--saveGraph</arg><arg>${saveGraph}</arg>
|
2020-04-27 10:48:31 +02:00
|
|
|
</spark>
|
|
|
|
<ok to="wait2"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
2020-05-08 12:33:31 +02:00
|
|
|
|
2020-04-27 10:48:31 +02:00
|
|
|
<action name="join_propagate_dataset">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Result2CommunitySemRelPropagation-Dataset</name>
|
2020-05-07 18:22:26 +02:00
|
|
|
<class>eu.dnetlib.dhp.resulttocommunityfromsemrel.SparkResultToCommunityThroughSemRelJob</class>
|
2020-05-12 15:49:28 +02:00
|
|
|
<jar>dhp-enrichment-${projectVersion}.jar</jar>
|
2020-04-27 10:48:31 +02:00
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${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}${spark2EventLogDir}
|
|
|
|
--conf spark.dynamicAllocation.enabled=true
|
|
|
|
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--preparedInfoPath</arg><arg>${workingDir}/preparedInfo/mergedCommunityAssoc</arg>
|
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}/dataset</arg>
|
|
|
|
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
|
|
|
|
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
|
2020-04-30 12:45:28 +02:00
|
|
|
<arg>--outputPath</arg><arg>${outputPath}/dataset</arg>
|
2020-05-05 10:03:32 +02:00
|
|
|
<arg>--saveGraph</arg><arg>${saveGraph}</arg>
|
2020-04-27 10:48:31 +02:00
|
|
|
</spark>
|
|
|
|
<ok to="wait2"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
2020-05-08 12:33:31 +02:00
|
|
|
|
2020-04-27 10:48:31 +02:00
|
|
|
<action name="join_propagate_otherresearchproduct">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Result2CommunitySemRelPropagation-ORP</name>
|
2020-05-07 18:22:26 +02:00
|
|
|
<class>eu.dnetlib.dhp.resulttocommunityfromsemrel.SparkResultToCommunityThroughSemRelJob</class>
|
2020-05-12 15:49:28 +02:00
|
|
|
<jar>dhp-enrichment-${projectVersion}.jar</jar>
|
2020-04-27 10:48:31 +02:00
|
|
|
<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.dynamicAllocation.enabled=true
|
|
|
|
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--preparedInfoPath</arg><arg>${workingDir}/preparedInfo/mergedCommunityAssoc</arg>
|
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}/otherresearchproduct</arg>
|
|
|
|
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
|
|
|
|
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
|
2020-04-30 12:45:28 +02:00
|
|
|
<arg>--outputPath</arg><arg>${outputPath}/otherresearchproduct</arg>
|
2020-05-05 10:03:32 +02:00
|
|
|
<arg>--saveGraph</arg><arg>${saveGraph}</arg>
|
2020-04-27 10:48:31 +02:00
|
|
|
</spark>
|
|
|
|
<ok to="wait2"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
2020-05-08 12:33:31 +02:00
|
|
|
|
2020-04-27 10:48:31 +02:00
|
|
|
<action name="join_propagate_software">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Result2CommunitySemRelPropagation-Software</name>
|
2020-05-07 18:22:26 +02:00
|
|
|
<class>eu.dnetlib.dhp.resulttocommunityfromsemrel.SparkResultToCommunityThroughSemRelJob</class>
|
2020-05-12 15:49:28 +02:00
|
|
|
<jar>dhp-enrichment-${projectVersion}.jar</jar>
|
2020-04-27 10:48:31 +02:00
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${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}${spark2EventLogDir}
|
|
|
|
--conf spark.dynamicAllocation.enabled=true
|
|
|
|
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--preparedInfoPath</arg><arg>${workingDir}/preparedInfo/mergedCommunityAssoc</arg>
|
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}/software</arg>
|
|
|
|
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
|
|
|
|
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
|
2020-04-30 12:45:28 +02:00
|
|
|
<arg>--outputPath</arg><arg>${outputPath}/software</arg>
|
2020-05-05 10:03:32 +02:00
|
|
|
<arg>--saveGraph</arg><arg>${saveGraph}</arg>
|
2020-04-27 10:48:31 +02:00
|
|
|
</spark>
|
|
|
|
<ok to="wait2"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<join name="wait2" to="End"/>
|
2020-02-28 18:26:04 +01:00
|
|
|
|
2020-05-08 12:33:31 +02:00
|
|
|
<end name="End"/>
|
|
|
|
|
2020-02-28 18:26:04 +01:00
|
|
|
</workflow-app>
|