2020-05-11 09:38:27 +02:00
|
|
|
<workflow-app name="import MAG into HDFS" xmlns="uri:oozie:workflow:0.5">
|
|
|
|
<parameters>
|
|
|
|
<property>
|
|
|
|
<name>sourcePath</name>
|
|
|
|
<description>the working dir base path</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>targetPath</name>
|
|
|
|
<description>the working dir base 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>
|
|
|
|
|
2020-05-28 09:57:46 +02:00
|
|
|
<start to="PreprocessMag"/>
|
2020-05-11 09:38:27 +02:00
|
|
|
|
|
|
|
|
|
|
|
<kill name="Kill">
|
|
|
|
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
|
|
</kill>
|
|
|
|
|
|
|
|
<action name="ResetWorkingPath">
|
|
|
|
<fs>
|
2020-05-28 09:57:46 +02:00
|
|
|
<delete path='${targetPath}/preprocess'/>
|
|
|
|
<mkdir path='${targetPath}/preprocess'/>
|
2020-05-11 09:38:27 +02:00
|
|
|
</fs>
|
2020-05-13 10:38:04 +02:00
|
|
|
<ok to="PreprocessMag"/>
|
2020-05-11 09:38:27 +02:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="ConvertMagToDataset">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Convert Mag to Dataset</name>
|
|
|
|
<class>eu.dnetlib.doiboost.mag.SparkImportMagIntoDataset</class>
|
|
|
|
<jar>dhp-doiboost-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--driver-memory=${sparkDriverMemory}
|
|
|
|
${sparkExtraOPT}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}</arg>
|
|
|
|
<arg>--targetPath</arg><arg>${targetPath}</arg>
|
|
|
|
<arg>--master</arg><arg>yarn-cluster</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="End"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-05-13 10:38:04 +02:00
|
|
|
|
|
|
|
|
|
|
|
<action name="PreprocessMag">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Convert Mag to Dataset</name>
|
|
|
|
<class>eu.dnetlib.doiboost.mag.SparkPreProcessMAG</class>
|
|
|
|
<jar>dhp-doiboost-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
|
|
|
--driver-memory=${sparkDriverMemory}
|
2020-05-19 09:24:45 +02:00
|
|
|
--conf spark.sql.shuffle.partitions=3840
|
2020-05-13 10:38:04 +02:00
|
|
|
${sparkExtraOPT}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}</arg>
|
|
|
|
<arg>--targetPath</arg><arg>${targetPath}</arg>
|
|
|
|
<arg>--master</arg><arg>yarn-cluster</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="End"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-05-11 09:38:27 +02:00
|
|
|
<end name="End"/>
|
|
|
|
</workflow-app>
|