2020-06-01 17:52:12 +02:00
|
|
|
<workflow-app name="Create DOIBoostActionSet" xmlns="uri:oozie:workflow:0.5">
|
2020-05-22 15:15:09 +02:00
|
|
|
<parameters>
|
|
|
|
<property>
|
2020-06-04 14:39:20 +02:00
|
|
|
<name>hostedByMapPath</name>
|
|
|
|
<description>the Hosted By Map Path</description>
|
2020-05-22 15:15:09 +02:00
|
|
|
</property>
|
2020-05-26 09:15:33 +02:00
|
|
|
<property>
|
|
|
|
<name>affiliationPath</name>
|
|
|
|
<description>the Affliation Path</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>paperAffiliationPath</name>
|
|
|
|
<description>the paperAffiliation Path</description>
|
|
|
|
</property>
|
2020-05-22 15:15:09 +02:00
|
|
|
<property>
|
|
|
|
<name>workingDirPath</name>
|
|
|
|
<description>the Working Path</description>
|
|
|
|
</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>
|
|
|
|
</parameters>
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-06-04 14:39:20 +02:00
|
|
|
<start to="GenerateActionSet"/>
|
2020-05-22 15:15:09 +02:00
|
|
|
|
|
|
|
|
|
|
|
<kill name="Kill">
|
|
|
|
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
|
|
</kill>
|
|
|
|
|
|
|
|
<action name="ResetWorkingPath">
|
|
|
|
<fs>
|
|
|
|
<delete path='${workingDirPath}'/>
|
|
|
|
<mkdir path='${workingDirPath}'/>
|
|
|
|
</fs>
|
|
|
|
<ok to="CreateDOIBoost"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="CreateDOIBoost">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Create DOIBoost Infospace</name>
|
|
|
|
<class>eu.dnetlib.doiboost.SparkGenerateDoiBoost</class>
|
|
|
|
<jar>dhp-doiboost-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--driver-memory=${sparkDriverMemory}
|
|
|
|
--conf spark.sql.shuffle.partitions=3840
|
|
|
|
${sparkExtraOPT}
|
|
|
|
</spark-opts>
|
2020-06-04 14:39:20 +02:00
|
|
|
<arg>--hostedByMapPath</arg><arg>${hostedByMapPath}</arg>
|
2020-05-26 09:15:33 +02:00
|
|
|
<arg>--affiliationPath</arg><arg>${affiliationPath}</arg>
|
|
|
|
<arg>--paperAffiliationPath</arg><arg>${paperAffiliationPath}</arg>
|
2020-05-22 15:15:09 +02:00
|
|
|
<arg>--workingDirPath</arg><arg>${workingDirPath}</arg>
|
|
|
|
<arg>--master</arg><arg>yarn-cluster</arg>
|
|
|
|
</spark>
|
2020-06-04 14:39:20 +02:00
|
|
|
<ok to="GenerateActionSet"/>
|
2020-05-22 15:15:09 +02:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
|
2020-05-26 09:15:33 +02:00
|
|
|
<action name="GenerateActionSet">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Generate DOIBoost ActionSet</name>
|
|
|
|
<class>eu.dnetlib.doiboost.SparkGenerateDOIBoostActionSet</class>
|
|
|
|
<jar>dhp-doiboost-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--driver-memory=${sparkDriverMemory}
|
|
|
|
--conf spark.sql.shuffle.partitions=3840
|
|
|
|
${sparkExtraOPT}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--dbPublicationPath</arg><arg>${workingDirPath}/doiBoostPublicationFiltered</arg>
|
|
|
|
<arg>--dbDatasetPath</arg><arg>${workingDirPath}/crossrefDataset</arg>
|
|
|
|
<arg>--crossRefRelation</arg><arg>/data/doiboost/input/crossref/relations</arg>
|
|
|
|
<arg>--dbaffiliationRelationPath</arg><arg>${workingDirPath}/doiBoostPublicationAffiliation</arg>
|
2020-06-04 14:39:20 +02:00
|
|
|
<arg>-do</arg><arg>${workingDirPath}/doiBoostOrganization</arg>
|
2020-05-26 09:15:33 +02:00
|
|
|
<arg>--targetPath</arg><arg>${workingDirPath}/actionDataSet</arg>
|
|
|
|
<arg>--master</arg><arg>yarn-cluster</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="End"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
2020-05-22 15:15:09 +02:00
|
|
|
|
|
|
|
<end name="End"/>
|
|
|
|
</workflow-app>
|