2020-02-24 10:15:55 +01:00
|
|
|
<workflow-app name="import_infospace_graph" xmlns="uri:oozie:workflow:0.5">
|
|
|
|
<parameters>
|
|
|
|
<property>
|
|
|
|
<name>workingDirPath</name>
|
|
|
|
<description>the source path</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>graphPath</name>
|
|
|
|
<description>the graph path</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>index</name>
|
|
|
|
<description>index name</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>
|
|
|
|
|
2020-03-08 17:00:19 +01:00
|
|
|
<property>
|
|
|
|
<name>idScholix</name>
|
|
|
|
<description>the </description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>idSummary</name>
|
|
|
|
<description>number of cores used by single executor</description>
|
|
|
|
</property>
|
|
|
|
|
|
|
|
|
2020-02-24 10:15:55 +01:00
|
|
|
</parameters>
|
|
|
|
|
2020-03-08 17:00:19 +01:00
|
|
|
<start to="indexScholix"/>
|
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>${nameNode}</name-node>
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>calculate for each ID the number of related Dataset, publication and Unknown</name>
|
|
|
|
<class>eu.dnetlib.dhp.provision.SparkExtractRelationCount</class>
|
|
|
|
<jar>dhp-graph-provision-${projectVersion}.jar</jar>
|
|
|
|
<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>${nameNode}</name-node>
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>generate Summary</name>
|
|
|
|
<class>eu.dnetlib.dhp.provision.SparkGenerateSummary</class>
|
|
|
|
<jar>dhp-graph-provision-${projectVersion}.jar</jar>
|
|
|
|
<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>${nameNode}</name-node>
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
<mode>cluster</mode>
|
2020-03-19 11:11:07 +01:00
|
|
|
<name>generate Scholix</name>
|
|
|
|
<class>eu.dnetlib.dhp.provision.SparkGenerateScholix</class>
|
2020-02-24 10:15:55 +01:00
|
|
|
<jar>dhp-graph-provision-${projectVersion}.jar</jar>
|
2020-03-19 11:11:07 +01:00
|
|
|
<spark-opts>--executor-memory 9G --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>${graphPath}</arg>
|
2020-02-24 10:15:55 +01:00
|
|
|
</spark>
|
2020-03-19 11:11:07 +01:00
|
|
|
<ok to="indexScholix"/>
|
2020-02-24 10:15:55 +01:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-03-19 11:11:07 +01:00
|
|
|
<action name="indexSummary">
|
2020-02-26 12:09:06 +01:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<job-tracker>${jobTracker}</job-tracker>
|
|
|
|
<name-node>${nameNode}</name-node>
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
<mode>cluster</mode>
|
2020-03-19 11:11:07 +01:00
|
|
|
<name>generate Summary</name>
|
|
|
|
<class>eu.dnetlib.dhp.provision.SparkIndexCollectionOnES</class>
|
2020-02-26 12:09:06 +01:00
|
|
|
<jar>dhp-graph-provision-${projectVersion}.jar</jar>
|
2020-03-19 11:11:07 +01:00
|
|
|
<spark-opts>--executor-memory ${sparkExecutorMemory} --driver-memory=${sparkDriverMemory} ${sparkExtraOPT} --conf spark.dynamicAllocation.maxExecutors="64" </spark-opts>
|
2020-02-26 12:09:06 +01:00
|
|
|
<arg>-mt</arg> <arg>yarn-cluster</arg>
|
2020-03-19 11:11:07 +01:00
|
|
|
<arg>--sourcePath</arg><arg>${workingDirPath}/summary</arg>
|
|
|
|
<arg>--index</arg><arg>${index}_object</arg>
|
|
|
|
<arg>--idPath</arg><arg>id</arg>
|
|
|
|
<arg>--type</arg><arg>summary</arg>
|
2020-02-26 12:09:06 +01:00
|
|
|
</spark>
|
2020-03-08 17:00:19 +01:00
|
|
|
<ok to="indexScholix"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="indexScholix">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<job-tracker>${jobTracker}</job-tracker>
|
|
|
|
<name-node>${nameNode}</name-node>
|
|
|
|
<master>yarn-cluster</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>index scholix</name>
|
|
|
|
<class>eu.dnetlib.dhp.provision.SparkIndexCollectionOnES</class>
|
|
|
|
<jar>dhp-graph-provision-${projectVersion}.jar</jar>
|
2020-03-19 11:11:07 +01:00
|
|
|
<spark-opts>--executor-memory ${sparkExecutorMemory} --driver-memory=${sparkDriverMemory} ${sparkExtraOPT} --conf spark.dynamicAllocation.maxExecutors="16" </spark-opts>
|
2020-03-08 17:00:19 +01:00
|
|
|
<arg>-mt</arg> <arg>yarn-cluster</arg>
|
2020-03-19 11:11:07 +01:00
|
|
|
<arg>--sourcePath</arg><arg>${workingDirPath}/scholix_json</arg>
|
2020-03-08 17:00:19 +01:00
|
|
|
<arg>--index</arg><arg>${index}_scholix</arg>
|
2020-03-19 11:11:07 +01:00
|
|
|
<arg>--idPath</arg><arg>identifier</arg>
|
|
|
|
<arg>--type</arg><arg>scholix</arg>
|
2020-03-08 17:00:19 +01:00
|
|
|
</spark>
|
2020-02-26 12:22:43 +01:00
|
|
|
<ok to="End"/>
|
2020-02-26 12:09:06 +01:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<end name="End"/>
|
2020-02-24 10:15:55 +01:00
|
|
|
</workflow-app>
|