271 lines
13 KiB
XML
271 lines
13 KiB
XML
<workflow-app name="fuse graph entities by id" xmlns="uri:oozie:workflow:0.5">
|
|
|
|
<parameters>
|
|
<property>
|
|
<name>graphInputPath</name>
|
|
<description>the graph root input path</description>
|
|
</property>
|
|
<property>
|
|
<name>outputPath</name>
|
|
<description>the graph root output 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>
|
|
<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>
|
|
<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>
|
|
</parameters>
|
|
|
|
<start to="fuse_graph"/>
|
|
|
|
<kill name="Kill">
|
|
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
</kill>
|
|
|
|
<action name="fuse_graph">
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
<master>yarn</master>
|
|
<mode>cluster</mode>
|
|
<name>Fuse graph entities by ID</name>
|
|
<class>eu.dnetlib.dhp.oa.graph.fuse.FuseGraphResultsSparkJob</class>
|
|
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
|
|
<spark-opts>
|
|
--executor-cores=${sparkExecutorCores}
|
|
--executor-memory=${sparkExecutorMemory}
|
|
--driver-memory=${sparkDriverMemory}
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
--conf spark.sql.shuffle.partitions=7680
|
|
</spark-opts>
|
|
<arg>--graphInputPath</arg><arg>${graphInputPath}</arg>
|
|
<arg>--outputPath</arg><arg>${workingDir}/entities</arg>
|
|
</spark>
|
|
<ok to="wait_merge"/>
|
|
<error to="Kill"/>
|
|
</action>
|
|
|
|
<action name="merge_dataset">
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
<master>yarn</master>
|
|
<mode>cluster</mode>
|
|
<name>Merge datasets</name>
|
|
<class>eu.dnetlib.dhp.oa.graph.merge.MergeGraphTableSparkJob</class>
|
|
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
|
|
<spark-opts>
|
|
--executor-cores=${sparkExecutorCores}
|
|
--executor-memory=${sparkExecutorMemory}
|
|
--driver-memory=${sparkDriverMemory}
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
--conf spark.sql.shuffle.partitions=7680
|
|
</spark-opts>
|
|
<arg>--betaInputPath</arg><arg>${betaInputGgraphPath}/dataset</arg>
|
|
<arg>--prodInputPath</arg><arg>${prodInputGgraphPath}/dataset</arg>
|
|
<arg>--outputPath</arg><arg>${graphOutputPath}/dataset</arg>
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
|
|
<arg>--priority</arg><arg>${priority}</arg>
|
|
</spark>
|
|
<ok to="wait_merge"/>
|
|
<error to="Kill"/>
|
|
</action>
|
|
|
|
<action name="merge_otherresearchproduct">
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
<master>yarn</master>
|
|
<mode>cluster</mode>
|
|
<name>Merge otherresearchproducts</name>
|
|
<class>eu.dnetlib.dhp.oa.graph.merge.MergeGraphTableSparkJob</class>
|
|
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
|
|
<spark-opts>
|
|
--executor-cores=${sparkExecutorCores}
|
|
--executor-memory=${sparkExecutorMemory}
|
|
--driver-memory=${sparkDriverMemory}
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
--conf spark.sql.shuffle.partitions=7680
|
|
</spark-opts>
|
|
<arg>--betaInputPath</arg><arg>${betaInputGgraphPath}/otherresearchproduct</arg>
|
|
<arg>--prodInputPath</arg><arg>${prodInputGgraphPath}/otherresearchproduct</arg>
|
|
<arg>--outputPath</arg><arg>${graphOutputPath}/otherresearchproduct</arg>
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
|
|
<arg>--priority</arg><arg>${priority}</arg>
|
|
</spark>
|
|
<ok to="wait_merge"/>
|
|
<error to="Kill"/>
|
|
</action>
|
|
|
|
<action name="merge_software">
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
<master>yarn</master>
|
|
<mode>cluster</mode>
|
|
<name>Merge softwares</name>
|
|
<class>eu.dnetlib.dhp.oa.graph.merge.MergeGraphTableSparkJob</class>
|
|
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
|
|
<spark-opts>
|
|
--executor-cores=${sparkExecutorCores}
|
|
--executor-memory=${sparkExecutorMemory}
|
|
--driver-memory=${sparkDriverMemory}
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
--conf spark.sql.shuffle.partitions=7680
|
|
</spark-opts>
|
|
<arg>--betaInputPath</arg><arg>${betaInputGgraphPath}/software</arg>
|
|
<arg>--prodInputPath</arg><arg>${prodInputGgraphPath}/software</arg>
|
|
<arg>--outputPath</arg><arg>${graphOutputPath}/software</arg>
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
|
|
<arg>--priority</arg><arg>${priority}</arg>
|
|
</spark>
|
|
<ok to="wait_merge"/>
|
|
<error to="Kill"/>
|
|
</action>
|
|
|
|
<action name="merge_datasource">
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
<master>yarn</master>
|
|
<mode>cluster</mode>
|
|
<name>Merge datasources</name>
|
|
<class>eu.dnetlib.dhp.oa.graph.merge.MergeGraphTableSparkJob</class>
|
|
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
|
|
<spark-opts>
|
|
--executor-cores=${sparkExecutorCores}
|
|
--executor-memory=${sparkExecutorMemory}
|
|
--driver-memory=${sparkDriverMemory}
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
--conf spark.sql.shuffle.partitions=7680
|
|
</spark-opts>
|
|
<arg>--betaInputPath</arg><arg>${betaInputGgraphPath}/datasource</arg>
|
|
<arg>--prodInputPath</arg><arg>${prodInputGgraphPath}/datasource</arg>
|
|
<arg>--outputPath</arg><arg>${graphOutputPath}/datasource</arg>
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Datasource</arg>
|
|
<arg>--priority</arg><arg>${priority}</arg>
|
|
</spark>
|
|
<ok to="wait_merge"/>
|
|
<error to="Kill"/>
|
|
</action>
|
|
|
|
<action name="merge_organization">
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
<master>yarn</master>
|
|
<mode>cluster</mode>
|
|
<name>Merge organizations</name>
|
|
<class>eu.dnetlib.dhp.oa.graph.merge.MergeGraphTableSparkJob</class>
|
|
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
|
|
<spark-opts>
|
|
--executor-cores=${sparkExecutorCores}
|
|
--executor-memory=${sparkExecutorMemory}
|
|
--driver-memory=${sparkDriverMemory}
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
--conf spark.sql.shuffle.partitions=7680
|
|
</spark-opts>
|
|
<arg>--betaInputPath</arg><arg>${betaInputGgraphPath}/organization</arg>
|
|
<arg>--prodInputPath</arg><arg>${prodInputGgraphPath}/organization</arg>
|
|
<arg>--outputPath</arg><arg>${graphOutputPath}/organization</arg>
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Organization</arg>
|
|
<arg>--priority</arg><arg>${priority}</arg>
|
|
</spark>
|
|
<ok to="wait_merge"/>
|
|
<error to="Kill"/>
|
|
</action>
|
|
|
|
<action name="merge_project">
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
<master>yarn</master>
|
|
<mode>cluster</mode>
|
|
<name>Merge projects</name>
|
|
<class>eu.dnetlib.dhp.oa.graph.merge.MergeGraphTableSparkJob</class>
|
|
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
|
|
<spark-opts>
|
|
--executor-cores=${sparkExecutorCores}
|
|
--executor-memory=${sparkExecutorMemory}
|
|
--driver-memory=${sparkDriverMemory}
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
--conf spark.sql.shuffle.partitions=7680
|
|
</spark-opts>
|
|
<arg>--betaInputPath</arg><arg>${betaInputGgraphPath}/project</arg>
|
|
<arg>--prodInputPath</arg><arg>${prodInputGgraphPath}/project</arg>
|
|
<arg>--outputPath</arg><arg>${graphOutputPath}/project</arg>
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Project</arg>
|
|
<arg>--priority</arg><arg>${priority}</arg>
|
|
</spark>
|
|
<ok to="wait_merge"/>
|
|
<error to="Kill"/>
|
|
</action>
|
|
|
|
<action name="merge_relation">
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
<master>yarn</master>
|
|
<mode>cluster</mode>
|
|
<name>Merge relations</name>
|
|
<class>eu.dnetlib.dhp.oa.graph.merge.MergeGraphTableSparkJob</class>
|
|
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
|
|
<spark-opts>
|
|
--executor-cores=${sparkExecutorCores}
|
|
--executor-memory=${sparkExecutorMemory}
|
|
--driver-memory=${sparkDriverMemory}
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
--conf spark.sql.shuffle.partitions=7680
|
|
</spark-opts>
|
|
<arg>--betaInputPath</arg><arg>${betaInputGgraphPath}/relation</arg>
|
|
<arg>--prodInputPath</arg><arg>${prodInputGgraphPath}/relation</arg>
|
|
<arg>--outputPath</arg><arg>${graphOutputPath}/relation</arg>
|
|
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Relation</arg>
|
|
<arg>--priority</arg><arg>${priority}</arg>
|
|
</spark>
|
|
<ok to="wait_merge"/>
|
|
<error to="Kill"/>
|
|
</action>
|
|
|
|
<join name="wait_merge" to="End"/>
|
|
|
|
<end name="End"/>
|
|
</workflow-app> |