2019-12-17 16:24:49 +01:00
|
|
|
<workflow-app name="index_infospace_graph" xmlns="uri:oozie:workflow:0.5">
|
2020-03-18 15:43:54 +01:00
|
|
|
|
2019-12-17 16:24:49 +01:00
|
|
|
<parameters>
|
2020-04-04 14:03:43 +02:00
|
|
|
<property>
|
|
|
|
<name>inputGraphRootPath</name>
|
|
|
|
<description>root location of input materialized graph</description>
|
|
|
|
</property>
|
|
|
|
|
2019-12-17 16:24:49 +01:00
|
|
|
<property>
|
2020-04-01 19:07:30 +02:00
|
|
|
<name>sparkDriverMemoryForJoining</name>
|
|
|
|
<description>memory for driver process</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>sparkExecutorMemoryForJoining</name>
|
|
|
|
<description>memory for individual executor</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>sparkExecutorCoresForJoining</name>
|
|
|
|
<description>number of cores used by single executor</description>
|
2019-12-17 16:24:49 +01:00
|
|
|
</property>
|
|
|
|
<property>
|
2020-04-01 19:07:30 +02:00
|
|
|
<name>sparkDriverMemoryForIndexing</name>
|
2019-12-17 16:24:49 +01:00
|
|
|
<description>memory for driver process</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
2020-04-01 19:07:30 +02:00
|
|
|
<name>sparkExecutorMemoryForIndexing</name>
|
2019-12-17 16:24:49 +01:00
|
|
|
<description>memory for individual executor</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
2020-04-01 19:07:30 +02:00
|
|
|
<name>sparkExecutorCoresForIndexing</name>
|
2019-12-17 16:24:49 +01:00
|
|
|
<description>number of cores used by single executor</description>
|
|
|
|
</property>
|
2020-04-02 18:44:09 +02:00
|
|
|
<property>
|
|
|
|
<name>oozieActionShareLibForSpark2</name>
|
|
|
|
<description>oozie action sharelib for spark 2.*</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>spark2ExtraListeners</name>
|
|
|
|
<value>com.cloudera.spark.lineage.NavigatorAppListener</value>
|
|
|
|
<description>spark 2.* extra listeners classname</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>spark2SqlQueryExecutionListeners</name>
|
|
|
|
<value>com.cloudera.spark.lineage.NavigatorQueryListener</value>
|
|
|
|
<description>spark 2.* sql query execution listeners classname</description>
|
|
|
|
</property>
|
2020-01-21 16:35:44 +01:00
|
|
|
<property>
|
|
|
|
<name>spark2YarnHistoryServerAddress</name>
|
|
|
|
<description>spark 2.* yarn history server address</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>spark2EventLogDir</name>
|
|
|
|
<description>spark 2.* event log dir location</description>
|
|
|
|
</property>
|
2019-12-17 16:24:49 +01:00
|
|
|
</parameters>
|
|
|
|
|
2020-03-18 15:43:54 +01:00
|
|
|
<global>
|
|
|
|
<job-tracker>${jobTracker}</job-tracker>
|
|
|
|
<name-node>${nameNode}</name-node>
|
|
|
|
<configuration>
|
|
|
|
<property>
|
2020-04-02 18:44:09 +02:00
|
|
|
<name>oozie.action.sharelib.for.spark</name>
|
|
|
|
<value>${oozieActionShareLibForSpark2}</value>
|
2020-03-18 15:43:54 +01:00
|
|
|
</property>
|
|
|
|
</configuration>
|
|
|
|
</global>
|
|
|
|
|
2020-02-13 16:53:27 +01:00
|
|
|
<start to="reuse_records"/>
|
|
|
|
|
|
|
|
<decision name="reuse_records">
|
|
|
|
<switch>
|
2020-04-04 14:03:43 +02:00
|
|
|
<case to="prepare_relations">${wf:conf('reuseRecords') eq false}</case>
|
2020-02-13 16:53:27 +01:00
|
|
|
<case to="to_solr_index">${wf:conf('reuseRecords') eq true}</case>
|
|
|
|
<default to="adjancency_lists"/>
|
|
|
|
</switch>
|
|
|
|
</decision>
|
2019-12-17 16:24:49 +01:00
|
|
|
|
|
|
|
<kill name="Kill">
|
|
|
|
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
|
|
</kill>
|
|
|
|
|
2020-04-04 14:03:43 +02:00
|
|
|
<action name="prepare_relations">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>PrepareRelations</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.provision.PrepareRelationsJob</class>
|
|
|
|
<jar>dhp-graph-provision-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCoresForJoining}
|
|
|
|
--executor-memory=${sparkExecutorMemoryForJoining}
|
|
|
|
--driver-memory=${sparkDriverMemoryForJoining}
|
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--inputRelationsPath</arg><arg>${inputGraphRootPath}/relation</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/relation</arg>
|
|
|
|
</spark>
|
2020-04-04 17:41:31 +02:00
|
|
|
<ok to="fork_join_related_entities"/>
|
2020-04-04 14:03:43 +02:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-04-04 17:41:31 +02:00
|
|
|
<fork name="fork_join_related_entities">
|
|
|
|
<path start="join_relation_publication"/>
|
|
|
|
<path start="join_relation_dataset"/>
|
|
|
|
<path start="join_relation_otherresearchproduct"/>
|
|
|
|
<path start="join_relation_software"/>
|
|
|
|
<path start="join_relation_datasource"/>
|
|
|
|
<path start="join_relation_organization"/>
|
|
|
|
<path start="join_relation_project"/>
|
|
|
|
</fork>
|
|
|
|
|
2020-04-04 14:03:43 +02:00
|
|
|
<action name="join_relation_publication">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Join[relation.target = publication.id]</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1</class>
|
|
|
|
<jar>dhp-graph-provision-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCoresForJoining}
|
|
|
|
--executor-memory=${sparkExecutorMemoryForJoining}
|
|
|
|
--driver-memory=${sparkDriverMemoryForJoining}
|
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.shuffle.partitions=3840
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--inputRelationsPath</arg><arg>${workingDir}/relations</arg>
|
|
|
|
<arg>--inputEntityPath</arg><arg>${inputGraphRootPath}/publication</arg>
|
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/join_partial</arg>
|
|
|
|
</spark>
|
2020-04-04 17:41:31 +02:00
|
|
|
<ok to="join_relation"/>
|
2020-04-04 14:03:43 +02:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="join_relation_dataset">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Join[relation.target = dataset.id]</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1</class>
|
|
|
|
<jar>dhp-graph-provision-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCoresForJoining}
|
|
|
|
--executor-memory=${sparkExecutorMemoryForJoining}
|
|
|
|
--driver-memory=${sparkDriverMemoryForJoining}
|
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.shuffle.partitions=3840
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--inputRelationsPath</arg><arg>${workingDir}/relations</arg>
|
|
|
|
<arg>--inputEntityPath</arg><arg>${inputGraphRootPath}/dataset</arg>
|
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/join_partial</arg>
|
|
|
|
</spark>
|
2020-04-04 17:41:31 +02:00
|
|
|
<ok to="join_relation"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="join_relation_otherresearchproduct">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Join[relation.target = otherresearchproduct.id]</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1</class>
|
|
|
|
<jar>dhp-graph-provision-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCoresForJoining}
|
|
|
|
--executor-memory=${sparkExecutorMemoryForJoining}
|
|
|
|
--driver-memory=${sparkDriverMemoryForJoining}
|
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.shuffle.partitions=3840
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--inputRelationsPath</arg><arg>${workingDir}/relations</arg>
|
|
|
|
<arg>--inputEntityPath</arg><arg>${inputGraphRootPath}/otherresearchproduct</arg>
|
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/join_partial</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="join_relation"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="join_relation_software">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Join[relation.target = software.id]</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1</class>
|
|
|
|
<jar>dhp-graph-provision-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCoresForJoining}
|
|
|
|
--executor-memory=${sparkExecutorMemoryForJoining}
|
|
|
|
--driver-memory=${sparkDriverMemoryForJoining}
|
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.shuffle.partitions=3840
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--inputRelationsPath</arg><arg>${workingDir}/relations</arg>
|
|
|
|
<arg>--inputEntityPath</arg><arg>${inputGraphRootPath}/software</arg>
|
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/join_partial</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="join_relation"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="join_relation_datasource">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Join[relation.target = datasource.id]</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1</class>
|
|
|
|
<jar>dhp-graph-provision-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCoresForJoining}
|
|
|
|
--executor-memory=${sparkExecutorMemoryForJoining}
|
|
|
|
--driver-memory=${sparkDriverMemoryForJoining}
|
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.shuffle.partitions=3840
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--inputRelationsPath</arg><arg>${workingDir}/relations</arg>
|
|
|
|
<arg>--inputEntityPath</arg><arg>${inputGraphRootPath}/datasource</arg>
|
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Datasource</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/join_partial</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="join_relation"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="join_relation_organization">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Join[relation.target = organization.id]</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1</class>
|
|
|
|
<jar>dhp-graph-provision-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCoresForJoining}
|
|
|
|
--executor-memory=${sparkExecutorMemoryForJoining}
|
|
|
|
--driver-memory=${sparkDriverMemoryForJoining}
|
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.shuffle.partitions=3840
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--inputRelationsPath</arg><arg>${workingDir}/relations</arg>
|
|
|
|
<arg>--inputEntityPath</arg><arg>${inputGraphRootPath}/organization</arg>
|
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Organization</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/join_partial</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="join_relation"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="join_relation_project">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>Join[relation.target = project.id]</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase1</class>
|
|
|
|
<jar>dhp-graph-provision-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCoresForJoining}
|
|
|
|
--executor-memory=${sparkExecutorMemoryForJoining}
|
|
|
|
--driver-memory=${sparkDriverMemoryForJoining}
|
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.shuffle.partitions=3840
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--inputRelationsPath</arg><arg>${workingDir}/relations</arg>
|
|
|
|
<arg>--inputEntityPath</arg><arg>${inputGraphRootPath}/project</arg>
|
|
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Project</arg>
|
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/join_partial</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="join_relation"/>
|
2020-04-04 14:03:43 +02:00
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-04-04 17:41:31 +02:00
|
|
|
<join name="join_relation" to="join_all_entities"/>
|
|
|
|
|
2020-04-04 14:03:43 +02:00
|
|
|
<action name="join_all_entities">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
2020-04-04 17:41:31 +02:00
|
|
|
<name>Join[entities.id = relatedEntity.source]</name>
|
2020-04-04 14:03:43 +02:00
|
|
|
<class>eu.dnetlib.dhp.oa.provision.CreateRelatedEntitiesJob_phase2</class>
|
|
|
|
<jar>dhp-graph-provision-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCoresForJoining}
|
|
|
|
--executor-memory=${sparkExecutorMemoryForJoining}
|
|
|
|
--driver-memory=${sparkDriverMemoryForJoining}
|
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.shuffle.partitions=3840
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--inputEntityPath</arg><arg>${inputGraphRootPath}</arg>
|
2020-04-04 17:41:31 +02:00
|
|
|
<arg>--inputRelatedEntitiesPath</arg><arg>${workingDir}/join_partial</arg>
|
2020-04-04 14:03:43 +02:00
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/join_entities</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="adjancency_lists"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2020-01-30 17:46:29 +01:00
|
|
|
<action name="adjancency_lists">
|
2019-12-17 16:24:49 +01:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
2020-02-13 16:53:27 +01:00
|
|
|
<master>yarn</master>
|
2019-12-17 16:24:49 +01:00
|
|
|
<mode>cluster</mode>
|
2020-01-30 17:46:29 +01:00
|
|
|
<name>build_adjacency_lists</name>
|
2020-04-04 14:03:43 +02:00
|
|
|
<class>eu.dnetlib.dhp.oa.provision.AdjacencyListBuilderJob</class>
|
|
|
|
<jar>dhp-graph-provision-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-cores=${sparkExecutorCoresForJoining}
|
|
|
|
--executor-memory=${sparkExecutorMemoryForJoining}
|
|
|
|
--driver-memory=${sparkDriverMemoryForJoining}
|
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.shuffle.partitions=3840
|
|
|
|
</spark-opts>
|
2020-04-04 17:41:31 +02:00
|
|
|
<arg>--inputPath</arg> <arg>${workingDir}/join_entities</arg>
|
2020-04-04 14:03:43 +02:00
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/joined</arg>
|
|
|
|
</spark>
|
|
|
|
<ok to="convert_to_xml"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="convert_to_xml">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>build_adjacency_lists</name>
|
|
|
|
<class>eu.dnetlib.dhp.oa.provision.XmlConverterJob</class>
|
2019-12-17 16:24:49 +01:00
|
|
|
<jar>dhp-graph-provision-${projectVersion}.jar</jar>
|
2020-01-21 16:35:44 +01:00
|
|
|
<spark-opts>
|
2020-04-02 18:44:09 +02:00
|
|
|
--executor-cores=${sparkExecutorCoresForJoining}
|
|
|
|
--executor-memory=${sparkExecutorMemoryForJoining}
|
2020-03-04 10:55:11 +01:00
|
|
|
--driver-memory=${sparkDriverMemoryForJoining}
|
2020-04-02 18:44:09 +02:00
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
2020-01-21 16:35:44 +01:00
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
2020-04-04 14:03:43 +02:00
|
|
|
--conf spark.sql.shuffle.partitions=3840
|
2020-01-21 16:35:44 +01:00
|
|
|
</spark-opts>
|
2020-04-04 17:41:31 +02:00
|
|
|
<arg>--inputPath</arg><arg>${workingDir}/joined</arg>
|
2020-04-04 14:03:43 +02:00
|
|
|
<arg>--outputPath</arg><arg>${workingDir}/xml</arg>
|
|
|
|
<arg>--isLookupUrl</arg><arg>${isLookupUrl}</arg>
|
|
|
|
<arg>--otherDsTypeId</arg><arg>${otherDsTypeId}</arg>
|
2019-12-17 16:24:49 +01:00
|
|
|
</spark>
|
2020-02-13 16:53:27 +01:00
|
|
|
<ok to="to_solr_index"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="to_solr_index">
|
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
|
|
|
<name>to_solr_index</name>
|
2020-03-27 10:42:17 +01:00
|
|
|
<class>eu.dnetlib.dhp.oa.provision.SparkXmlIndexingJob</class>
|
2020-02-13 16:53:27 +01:00
|
|
|
<jar>dhp-graph-provision-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
2020-03-04 10:55:11 +01:00
|
|
|
--executor-cores ${sparkExecutorCoresForIndexing}
|
|
|
|
--executor-memory ${sparkExecutorMemoryForIndexing}
|
|
|
|
--driver-memory=${sparkDriverMemoryForIndexing}
|
2020-02-14 12:28:52 +01:00
|
|
|
--conf spark.dynamicAllocation.maxExecutors=${sparkExecutorCoresForIndexing}
|
2020-04-02 18:44:09 +02:00
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
2020-02-13 16:53:27 +01:00
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
</spark-opts>
|
2020-04-04 14:03:43 +02:00
|
|
|
<arg>--inputPath</arg><arg>${workingDir}/xml</arg>
|
2020-04-04 17:41:31 +02:00
|
|
|
<arg>--isLookupUrl</arg> <arg>${isLookupUrl}</arg>
|
2020-02-13 16:53:27 +01:00
|
|
|
<arg>--format</arg><arg>${format}</arg>
|
|
|
|
<arg>--batchSize</arg><arg>${batchSize}</arg>
|
|
|
|
</spark>
|
2019-12-17 16:24:49 +01:00
|
|
|
<ok to="End"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<end name="End"/>
|
|
|
|
</workflow-app>
|