2020-03-26 09:49:23 +01:00
|
|
|
<workflow-app name="Materialize and Index graph to ElasticSearch" xmlns="uri:oozie:workflow:0.5">
|
2020-02-24 10:15:55 +01:00
|
|
|
<parameters>
|
|
|
|
<property>
|
|
|
|
<name>workingDirPath</name>
|
|
|
|
<description>the source path</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>graphPath</name>
|
|
|
|
<description>the graph 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>
|
2020-04-08 13:18:20 +02:00
|
|
|
<!-- <property>-->
|
|
|
|
<!-- <name>index</name>-->
|
|
|
|
<!-- <description>index name</description>-->
|
|
|
|
<!-- </property>-->
|
|
|
|
<!-- <property>-->
|
|
|
|
<!-- <name>idScholix</name>-->
|
|
|
|
<!-- <description>the identifier name of the scholix </description>-->
|
|
|
|
<!-- </property>-->
|
|
|
|
<!-- <property>-->
|
|
|
|
<!-- <name>idSummary</name>-->
|
|
|
|
<!-- <description>the identifier name of the summary</description>-->
|
|
|
|
<!-- </property>-->
|
2020-02-24 10:15:55 +01:00
|
|
|
</parameters>
|
|
|
|
|
2020-04-08 13:18:20 +02:00
|
|
|
<start to="DeleteTargetPath"/>
|
2020-02-24 10:15:55 +01:00
|
|
|
|
|
|
|
<kill name="Kill">
|
|
|
|
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
|
|
</kill>
|
|
|
|
<action name="DeleteTargetPath">
|
|
|
|
<fs>
|
|
|
|
<delete path='${workingDirPath}'/>
|
|
|
|
<mkdir path='${workingDirPath}'/>
|
|
|
|
</fs>
|
|
|
|
<ok to="CalculateRelatedItem"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="CalculateRelatedItem">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<job-tracker>${jobTracker}</job-tracker>
|
|
|
|
<name-node>${nameNode}</name-node>
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>calculate for each ID the number of related Dataset, publication and Unknown</name>
|
|
|
|
<class>eu.dnetlib.dhp.provision.SparkExtractRelationCount</class>
|
2020-03-31 14:41:35 +02:00
|
|
|
<jar>dhp-graph-provision-scholexplorer-${projectVersion}.jar</jar>
|
2020-02-24 10:15:55 +01:00
|
|
|
<spark-opts>--executor-memory ${sparkExecutorMemory} --driver-memory=${sparkDriverMemory} ${sparkExtraOPT}</spark-opts>
|
|
|
|
<arg>-mt</arg> <arg>yarn-cluster</arg>
|
|
|
|
<arg>--workingDirPath</arg><arg>${workingDirPath}</arg>
|
|
|
|
<arg>--relationPath</arg><arg>${graphPath}/relation</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="generateSummary"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="generateSummary">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<job-tracker>${jobTracker}</job-tracker>
|
|
|
|
<name-node>${nameNode}</name-node>
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>generate Summary</name>
|
|
|
|
<class>eu.dnetlib.dhp.provision.SparkGenerateSummary</class>
|
2020-03-31 14:41:35 +02:00
|
|
|
<jar>dhp-graph-provision-scholexplorer-${projectVersion}.jar</jar>
|
2020-02-24 10:15:55 +01:00
|
|
|
<spark-opts>--executor-memory ${sparkExecutorMemory} --driver-memory=${sparkDriverMemory} ${sparkExtraOPT}</spark-opts>
|
|
|
|
<arg>-mt</arg> <arg>yarn-cluster</arg>
|
|
|
|
<arg>--workingDirPath</arg><arg>${workingDirPath}</arg>
|
|
|
|
<arg>--graphPath</arg><arg>${graphPath}</arg>
|
|
|
|
</spark>
|
2020-03-19 11:11:07 +01:00
|
|
|
<ok to="generateScholix"/>
|
2020-02-24 10:15:55 +01:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-03-19 11:11:07 +01:00
|
|
|
<action name="generateScholix">
|
2020-02-24 10:15:55 +01:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<job-tracker>${jobTracker}</job-tracker>
|
|
|
|
<name-node>${nameNode}</name-node>
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
<mode>cluster</mode>
|
2020-03-19 11:11:07 +01:00
|
|
|
<name>generate Scholix</name>
|
|
|
|
<class>eu.dnetlib.dhp.provision.SparkGenerateScholix</class>
|
2020-03-31 14:41:35 +02:00
|
|
|
<jar>dhp-graph-provision-scholexplorer-${projectVersion}.jar</jar>
|
2020-03-26 09:40:50 +01:00
|
|
|
<spark-opts>--executor-memory 6G --driver-memory=${sparkDriverMemory} ${sparkExtraOPT}</spark-opts>
|
2020-02-24 10:15:55 +01:00
|
|
|
<arg>-mt</arg> <arg>yarn-cluster</arg>
|
2020-03-19 11:11:07 +01:00
|
|
|
<arg>--workingDirPath</arg><arg>${workingDirPath}</arg>
|
|
|
|
<arg>--graphPath</arg><arg>${graphPath}</arg>
|
2020-02-24 10:15:55 +01:00
|
|
|
</spark>
|
2020-03-26 09:40:50 +01:00
|
|
|
<ok to="End"/>
|
2020-02-24 10:15:55 +01:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-04-08 13:18:20 +02:00
|
|
|
<!-- <action name="indexSummary">-->
|
|
|
|
<!-- <spark xmlns="uri:oozie:spark-action:0.2">-->
|
|
|
|
<!-- <job-tracker>${jobTracker}</job-tracker>-->
|
|
|
|
<!-- <name-node>${nameNode}</name-node>-->
|
|
|
|
<!-- <master>yarn-cluster</master>-->
|
|
|
|
<!-- <mode>cluster</mode>-->
|
|
|
|
<!-- <name>index Summary</name>-->
|
|
|
|
<!-- <class>eu.dnetlib.dhp.provision.SparkIndexCollectionOnES</class>-->
|
|
|
|
<!-- <jar>dhp-graph-provision-scholexplorer-${projectVersion}.jar</jar>-->
|
|
|
|
<!-- <spark-opts>--executor-memory ${sparkExecutorMemory} --driver-memory=${sparkDriverMemory} ${sparkExtraOPT} --conf spark.dynamicAllocation.maxExecutors="32" </spark-opts>-->
|
|
|
|
<!-- <arg>-mt</arg> <arg>yarn-cluster</arg>-->
|
|
|
|
<!-- <arg>--sourcePath</arg><arg>${workingDirPath}/summary</arg>-->
|
|
|
|
<!-- <arg>--index</arg><arg>${index}_object</arg>-->
|
|
|
|
<!-- <arg>--idPath</arg><arg>id</arg>-->
|
|
|
|
<!-- <arg>--type</arg><arg>summary</arg>-->
|
|
|
|
<!-- </spark>-->
|
|
|
|
<!-- <ok to="indexScholix"/>-->
|
|
|
|
<!-- <error to="Kill"/>-->
|
|
|
|
<!-- </action>-->
|
2020-03-08 17:00:19 +01:00
|
|
|
|
2020-04-08 13:18:20 +02:00
|
|
|
<!-- <action name="indexScholix">-->
|
|
|
|
<!-- <spark xmlns="uri:oozie:spark-action:0.2">-->
|
|
|
|
<!-- <job-tracker>${jobTracker}</job-tracker>-->
|
|
|
|
<!-- <name-node>${nameNode}</name-node>-->
|
|
|
|
<!-- <master>yarn-cluster</master>-->
|
|
|
|
<!-- <mode>cluster</mode>-->
|
|
|
|
<!-- <name>index scholix</name>-->
|
|
|
|
<!-- <class>eu.dnetlib.dhp.provision.SparkIndexCollectionOnES</class>-->
|
|
|
|
<!-- <jar>dhp-graph-provision-scholexplorer-${projectVersion}.jar</jar>-->
|
|
|
|
<!-- <spark-opts>--executor-memory ${sparkExecutorMemory} --driver-memory=${sparkDriverMemory} ${sparkExtraOPT} --conf spark.dynamicAllocation.maxExecutors="8" </spark-opts>-->
|
|
|
|
<!-- <arg>-mt</arg> <arg>yarn-cluster</arg>-->
|
|
|
|
<!-- <arg>--sourcePath</arg><arg>${workingDirPath}/scholix_json</arg>-->
|
|
|
|
<!-- <arg>--index</arg><arg>${index}_scholix</arg>-->
|
|
|
|
<!-- <arg>--idPath</arg><arg>identifier</arg>-->
|
|
|
|
<!-- <arg>--type</arg><arg>scholix</arg>-->
|
|
|
|
<!-- </spark>-->
|
|
|
|
<!-- <ok to="End"/>-->
|
|
|
|
<!-- <error to="Kill"/>-->
|
|
|
|
<!-- </action>-->
|
2020-02-26 12:09:06 +01:00
|
|
|
|
|
|
|
<end name="End"/>
|
2020-02-24 10:15:55 +01:00
|
|
|
</workflow-app>
|