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>
|
2020-04-30 14:38:54 +02:00
|
|
|
<name>outputPath</name>
|
|
|
|
<description>the output path</description>
|
2020-03-02 17:03:31 +01:00
|
|
|
</property>
|
2023-12-05 09:13:33 +01:00
|
|
|
<property>
|
|
|
|
<name>baseURL</name>
|
|
|
|
<description>the community API base URL</description>
|
|
|
|
</property>
|
2020-03-02 17:03:31 +01:00
|
|
|
</parameters>
|
|
|
|
|
2020-05-14 10:25:41 +02:00
|
|
|
<global>
|
|
|
|
<job-tracker>${jobTracker}</job-tracker>
|
|
|
|
<name-node>${nameNode}</name-node>
|
|
|
|
<configuration>
|
|
|
|
<property>
|
|
|
|
<name>oozie.action.sharelib.for.spark</name>
|
|
|
|
<value>${oozieActionShareLibForSpark2}</value>
|
|
|
|
</property>
|
|
|
|
</configuration>
|
|
|
|
</global>
|
|
|
|
|
2023-10-16 11:26:07 +02:00
|
|
|
<start to="prepare_result_communitylist"/>
|
2020-03-02 17:03:31 +01:00
|
|
|
|
|
|
|
<kill name="Kill">
|
|
|
|
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
|
|
</kill>
|
|
|
|
|
2020-05-08 12:33:31 +02:00
|
|
|
<action name="reset_outputpath">
|
2020-04-23 12:13:34 +02:00
|
|
|
<fs>
|
2020-05-08 12:33:31 +02:00
|
|
|
<delete path="${outputPath}"/>
|
|
|
|
<mkdir path="${outputPath}"/>
|
2020-04-23 12:13:34 +02:00
|
|
|
</fs>
|
2020-04-30 14:38:54 +02:00
|
|
|
<ok to="copy_entities"/>
|
2020-04-23 12:13:34 +02:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-04-30 14:38:54 +02:00
|
|
|
<fork name="copy_entities">
|
|
|
|
<path start="copy_relation"/>
|
|
|
|
<path start="copy_organization"/>
|
|
|
|
<path start="copy_projects"/>
|
|
|
|
<path start="copy_datasources"/>
|
|
|
|
</fork>
|
|
|
|
|
2020-04-23 12:13:34 +02:00
|
|
|
<action name="copy_relation">
|
|
|
|
<distcp xmlns="uri:oozie:distcp-action:0.2">
|
|
|
|
<arg>${nameNode}/${sourcePath}/relation</arg>
|
2020-04-30 14:38:54 +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">
|
|
|
|
<arg>${nameNode}/${sourcePath}/organization</arg>
|
|
|
|
<arg>${nameNode}/${outputPath}/organization</arg>
|
|
|
|
</distcp>
|
|
|
|
<ok to="copy_wait"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
2020-05-08 12:33:31 +02:00
|
|
|
|
2020-04-30 14:38:54 +02:00
|
|
|
<action name="copy_projects">
|
|
|
|
<distcp xmlns="uri:oozie:distcp-action:0.2">
|
|
|
|
<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">
|
|
|
|
<arg>${nameNode}/${sourcePath}/datasource</arg>
|
|
|
|
<arg>${nameNode}/${outputPath}/datasource</arg>
|
2020-04-23 12:13:34 +02:00
|
|
|
</distcp>
|
2020-04-30 14:38:54 +02:00
|
|
|
<ok to="copy_wait"/>
|
2020-04-23 12:13:34 +02:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
2020-05-08 12:33:31 +02:00
|
|
|
|
2020-04-30 14:38:54 +02:00
|
|
|
<join name="copy_wait" to="prepare_result_communitylist"/>
|
2020-04-23 12:13:34 +02:00
|
|
|
|
|
|
|
<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>
|
2020-05-12 15:49:28 +02:00
|
|
|
<jar>dhp-enrichment-${projectVersion}.jar</jar>
|
2020-04-23 12:13:34 +02:00
|
|
|
<spark-opts>
|
2023-10-16 11:46:12 +02:00
|
|
|
--executor-cores=6
|
|
|
|
--executor-memory=5G
|
2023-10-20 10:14:20 +02:00
|
|
|
--conf spark.executor.memoryOverhead=3g
|
2023-10-16 11:46:12 +02:00
|
|
|
--conf spark.sql.shuffle.partitions=3284
|
2020-03-02 17:03:31 +01:00
|
|
|
--driver-memory=${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}${spark2EventLogDir}
|
|
|
|
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
|
2020-03-02 17:03:31 +01:00
|
|
|
</spark-opts>
|
2020-04-23 12:13:34 +02:00
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}/relation</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/preparedInfo/resultCommunityList</arg>
|
2020-05-14 10:25:41 +02:00
|
|
|
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
|
2023-12-05 09:13:33 +01:00
|
|
|
<arg>--baseURL</arg><arg>${baseURL}</arg>
|
2020-03-02 17:03:31 +01:00
|
|
|
</spark>
|
2023-10-16 11:26:07 +02:00
|
|
|
<ok to="exec-propagation"/>
|
2020-03-02 17:03:31 +01:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2023-10-16 11:26:07 +02:00
|
|
|
<action name="exec-propagation">
|
2020-04-23 12:13:34 +02:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
2023-12-05 09:13:33 +01:00
|
|
|
<name>community2resultfromorganization</name>
|
2020-05-07 18:22:26 +02:00
|
|
|
<class>eu.dnetlib.dhp.resulttocommunityfromorganization.SparkResultToCommunityFromOrganizationJob</class>
|
2020-05-12 15:49:28 +02:00
|
|
|
<jar>dhp-enrichment-${projectVersion}.jar</jar>
|
2020-04-23 12:13:34 +02:00
|
|
|
<spark-opts>
|
2023-10-16 11:46:12 +02:00
|
|
|
--executor-cores=6
|
|
|
|
--executor-memory=5G
|
2023-10-20 10:14:20 +02:00
|
|
|
--conf spark.executor.memoryOverhead=3g
|
2023-10-16 11:46:12 +02:00
|
|
|
--conf spark.sql.shuffle.partitions=3284
|
2020-04-23 12:13:34 +02:00
|
|
|
--driver-memory=${sparkDriverMemory}
|
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
--conf spark.dynamicAllocation.maxExecutors=${spark2MaxExecutors}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--preparedInfoPath</arg><arg>${workingDir}/preparedInfo/resultCommunityList</arg>
|
2023-10-16 11:26:07 +02:00
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}/</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${outputPath}/</arg>
|
2020-04-23 12:13:34 +02:00
|
|
|
</spark>
|
2023-10-16 11:26:07 +02:00
|
|
|
<ok to="End"/>
|
2020-04-23 12:13:34 +02:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
2020-05-08 12:33:31 +02:00
|
|
|
|
2020-04-23 12:13:34 +02:00
|
|
|
|
|
|
|
|
2020-03-02 17:03:31 +01:00
|
|
|
<end name="End"/>
|
2020-05-08 12:33:31 +02:00
|
|
|
|
2020-03-02 17:03:31 +01:00
|
|
|
</workflow-app>
|