changed to split in two steps the generation of the crossref dataset

This commit is contained in:
Miriam Baglioni 2021-06-18 09:41:23 +02:00
parent 95885bcf12
commit 3585e53da3
1 changed files with 30 additions and 3 deletions

View File

@ -142,6 +142,32 @@
<arg>--workingPath</arg><arg>${crossrefDumpPath}</arg>
<arg>--outputPath</arg><arg>${crossrefDumpPath}/files/</arg>
</java>
<ok to="UnpackCrossrefEntries"/>
<error to="Kill"/>
</action>
<action name="UnpackCrossrefEntries">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>SparkUnpackCrossrefEntries</name>
<class>eu.dnetlib.doiboost.crossref.UnpackCrtossrefEntries</class>
<jar>dhp-doiboost-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.sql.shuffle.partitions=3840
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--master</arg><arg>yarn-cluster</arg>
<arg>--sourcePath</arg><arg>${crossrefDumpPath}/files</arg>
<arg>--targetPath</arg><arg>${crossrefDumpPath}/crossref_unpack/</arg>
</spark>
<ok to="GenerateCrossrefDataset"/>
<error to="Kill"/>
</action>
@ -155,7 +181,7 @@
<jar>dhp-doiboost-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=7G
--executor-cores=4
--executor-cores=2
--driver-memory=7G
--conf spark.sql.shuffle.partitions=3840
--conf spark.extraListeners=${spark2ExtraListeners}
@ -164,7 +190,7 @@
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--master</arg><arg>yarn-cluster</arg>
<arg>--sourcePath</arg><arg>${crossrefDumpPath}/files/</arg>
<arg>--sourcePath</arg><arg>${crossrefDumpPath}/crossref_unpack/</arg>
<arg>--targetPath</arg><arg>${inputPathCrossref}/crossref_ds</arg>
</spark>
@ -174,7 +200,8 @@
<action name="removeFiles">
<fs>
<delete path="${crossrefDumpPath}/files"/>
<!-- <delete path="${crossrefDumpPath}/files"/>-->
<delete path="${crossrefDumpPath}/crossref_unpack/"/>
</fs>
<ok to="ResetMagWorkingPath"/>
<error to="Kill"/>