2020-04-21 16:03:51 +02:00
|
|
|
<workflow-app name="bulk_tagging" 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>pathMap</name>
|
|
|
|
<description>the json path associated to each selection field</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>outputPath</name>
|
|
|
|
<description>the output path</description>
|
|
|
|
</property>
|
2022-07-22 14:36:02 +02:00
|
|
|
|
|
|
|
|
|
|
|
<property>
|
|
|
|
<name>postgresURL</name>
|
|
|
|
<description>the url of the postgress server to query</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>postgresUser</name>
|
|
|
|
<description>the username to access the postgres db</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>postgresPassword</name>
|
|
|
|
<description>the postgres password</description>
|
|
|
|
</property>
|
|
|
|
|
2020-05-08 12:33:31 +02:00
|
|
|
</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>${oozieActionShareLibForSpark2}</value>
|
|
|
|
</property>
|
|
|
|
</configuration>
|
|
|
|
</global>
|
|
|
|
|
2023-10-11 18:17:35 +02:00
|
|
|
<start to="exec_bulktag"/>
|
2020-05-08 12:33:31 +02:00
|
|
|
|
|
|
|
<kill name="Kill">
|
|
|
|
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
|
|
</kill>
|
|
|
|
|
|
|
|
<action name="reset_outputpath">
|
2020-04-21 16:03:51 +02:00
|
|
|
<fs>
|
2020-05-08 12:33:31 +02:00
|
|
|
<delete path="${outputPath}"/>
|
|
|
|
<mkdir path="${outputPath}"/>
|
2020-04-21 16:03:51 +02:00
|
|
|
</fs>
|
2020-04-30 14:38:54 +02:00
|
|
|
<ok to="copy_entities"/>
|
2020-04-21 16:03:51 +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>
|
|
|
|
|
2020-04-21 16:03:51 +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>
|
2020-04-21 16:03:51 +02:00
|
|
|
</distcp>
|
2020-04-30 14:38:54 +02:00
|
|
|
<ok to="copy_wait"/>
|
2020-04-21 16:03:51 +02:00
|
|
|
<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>
|
|
|
|
</distcp>
|
|
|
|
<ok to="copy_wait"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2023-10-11 18:17:35 +02:00
|
|
|
<join name="copy_wait" to="exec_bulktag"/>
|
2020-04-30 14:38:54 +02:00
|
|
|
|
2023-10-11 18:17:35 +02:00
|
|
|
<action name="exec_bulktag">
|
2020-04-21 16:03:51 +02:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
<mode>cluster</mode>
|
2020-04-24 10:47:43 +02:00
|
|
|
<name>bulkTagging-publication</name>
|
2020-05-07 18:22:26 +02:00
|
|
|
<class>eu.dnetlib.dhp.bulktag.SparkBulkTagJob</class>
|
2020-05-12 15:49:28 +02:00
|
|
|
<jar>dhp-enrichment-${projectVersion}.jar</jar>
|
2020-04-21 16:03:51 +02:00
|
|
|
<spark-opts>
|
|
|
|
--num-executors=${sparkExecutorNumber}
|
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--driver-memory=${sparkDriverMemory}
|
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
</spark-opts>
|
2023-10-11 18:17:35 +02:00
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}/</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${outputPath}/</arg>
|
2020-05-04 13:58:52 +02:00
|
|
|
<arg>--pathMap</arg><arg>${pathMap}</arg>
|
2023-10-11 18:17:35 +02:00
|
|
|
<arg>--production</arg><arg>${production}</arg>
|
2020-04-21 16:03:51 +02:00
|
|
|
</spark>
|
2023-10-11 18:17:35 +02:00
|
|
|
<ok to="End"/>
|
2020-04-21 16:03:51 +02:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
2020-05-08 12:33:31 +02:00
|
|
|
|
2022-04-22 10:44:17 +02:00
|
|
|
|
2022-07-22 14:36:02 +02:00
|
|
|
|
2020-05-08 12:33:31 +02:00
|
|
|
<end name="End"/>
|
|
|
|
|
2020-04-08 10:31:29 +02:00
|
|
|
</workflow-app>
|