2020-03-02 17:03:31 +01:00
|
|
|
<workflow-app name="community_to_result_propagation" xmlns="uri:oozie:workflow:0.5">
|
|
|
|
<parameters>
|
|
|
|
<property>
|
|
|
|
<name>sourcePath</name>
|
|
|
|
<description>the source path</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>organizationtoresultcommunitymap</name>
|
|
|
|
<description>organization community map</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
2020-04-23 12:13:34 +02:00
|
|
|
<name>writeUpdate</name>
|
|
|
|
<description>writes the information found for the update. No double check done if the information is already present</description>
|
2020-03-02 17:03:31 +01:00
|
|
|
</property>
|
|
|
|
<property>
|
2020-04-23 12:13:34 +02:00
|
|
|
<name>saveGraph</name>
|
|
|
|
<description>writes new version of the graph after the propagation step</description>
|
2020-03-02 17:03:31 +01:00
|
|
|
</property>
|
|
|
|
</parameters>
|
|
|
|
|
2020-04-23 12:13:34 +02:00
|
|
|
<start to="reset-outputpath"/>
|
2020-03-02 17:03:31 +01:00
|
|
|
|
|
|
|
<kill name="Kill">
|
|
|
|
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
|
|
</kill>
|
|
|
|
|
2020-04-23 12:13:34 +02:00
|
|
|
<action name="reset-outputpath">
|
|
|
|
<fs>
|
|
|
|
<delete path='${workingDir}/resulttocommunityorganization_propagation'/>
|
|
|
|
</fs>
|
|
|
|
<ok to="copy_relation"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="copy_relation">
|
|
|
|
<distcp xmlns="uri:oozie:distcp-action:0.2">
|
2020-03-02 17:03:31 +01:00
|
|
|
<job-tracker>${jobTracker}</job-tracker>
|
|
|
|
<name-node>${nameNode}</name-node>
|
2020-04-23 12:13:34 +02:00
|
|
|
<arg>${nameNode}/${sourcePath}/relation</arg>
|
|
|
|
<arg>${nameNode}/${workingDir}/projecttoresult_propagation/relation</arg>
|
|
|
|
</distcp>
|
|
|
|
<ok to="prepare_result_communitylist"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="prepare_result_communitylist">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
2020-03-02 17:03:31 +01:00
|
|
|
<mode>cluster</mode>
|
2020-04-23 12:13:34 +02:00
|
|
|
<name>Prepare-Community-Result-Organization</name>
|
|
|
|
<class>eu.dnetlib.dhp.resulttocommunityfromorganization.PrepareResultCommunitySet</class>
|
2020-03-02 17:03:31 +01:00
|
|
|
<jar>dhp-propagation-${projectVersion}.jar</jar>
|
2020-04-23 12:13:34 +02:00
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--executor-memory=${sparkExecutorMemory}
|
2020-03-02 17:03:31 +01:00
|
|
|
--driver-memory=${sparkDriverMemory}
|
2020-04-23 12:13:34 +02:00
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
--conf spark.dynamicAllocation.enabled=true
|
|
|
|
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
|
2020-03-02 17:03:31 +01:00
|
|
|
</spark-opts>
|
2020-04-23 12:13:34 +02:00
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}/relation</arg>
|
|
|
|
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/preparedInfo/resultCommunityList</arg>
|
2020-03-02 17:03:31 +01:00
|
|
|
<arg>--organizationtoresultcommunitymap</arg><arg>${organizationtoresultcommunitymap}</arg>
|
|
|
|
</spark>
|
2020-04-23 12:13:34 +02:00
|
|
|
<ok to="fork-join-exec-propagation"/>
|
|
|
|
<!-- <ok to="End"/>-->
|
2020-03-02 17:03:31 +01:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-04-23 12:13:34 +02:00
|
|
|
<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>community2resultfromorganization-Publication</name>
|
|
|
|
<class>eu.dnetlib.dhp.resulttocommunityfromorganization.SparkResultToCommunityFromOrganizationJob2</class>
|
|
|
|
<jar>dhp-propagation-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${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}${spark2EventLogDir}
|
|
|
|
--conf spark.dynamicAllocation.enabled=true
|
|
|
|
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--preparedInfoPath</arg><arg>${workingDir}/preparedInfo/resultCommunityList</arg>
|
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}/publication</arg>
|
|
|
|
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
|
|
|
|
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/publication</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="wait2"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
<action name="join_propagate_dataset">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>community2resultfromorganization-Dataset</name>
|
|
|
|
<class>eu.dnetlib.dhp.resulttocommunityfromorganization.SparkResultToCommunityFromOrganizationJob2</class>
|
|
|
|
<jar>dhp-propagation-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${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}${spark2EventLogDir}
|
|
|
|
--conf spark.dynamicAllocation.enabled=true
|
|
|
|
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--preparedInfoPath</arg><arg>${workingDir}/preparedInfo/resultCommunityList</arg>
|
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}/dataset</arg>
|
|
|
|
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
|
|
|
|
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/dataset</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="wait2"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
<action name="join_propagate_otherresearchproduct">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>community2resultfromorganization-ORP</name>
|
|
|
|
<class>eu.dnetlib.dhp.resulttocommunityfromorganization.SparkResultToCommunityFromOrganizationJob2</class>
|
|
|
|
<jar>dhp-propagation-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${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}${spark2EventLogDir}
|
|
|
|
--conf spark.dynamicAllocation.enabled=true
|
|
|
|
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--preparedInfoPath</arg><arg>${workingDir}/preparedInfo/resultCommunityList</arg>
|
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}/otherresearchproduct</arg>
|
|
|
|
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
|
|
|
|
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/otherresearchproduct</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="wait2"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
<action name="join_propagate_software">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>community2resultfromorganization-Software</name>
|
|
|
|
<class>eu.dnetlib.dhp.resulttocommunityfromorganization.SparkResultToCommunityFromOrganizationJob2</class>
|
|
|
|
<jar>dhp-propagation-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${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}${spark2EventLogDir}
|
|
|
|
--conf spark.dynamicAllocation.enabled=true
|
|
|
|
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--preparedInfoPath</arg><arg>${workingDir}/preparedInfo/resultCommunityList</arg>
|
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}/software</arg>
|
|
|
|
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
|
|
|
|
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/software</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="wait2"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<join name="wait2" to="End"/>
|
|
|
|
|
2020-03-02 17:03:31 +01:00
|
|
|
<end name="End"/>
|
|
|
|
</workflow-app>
|