2020-03-26 19:44:19 +01:00
|
|
|
<workflow-app name="import_graph_as_hive_DB" xmlns="uri:oozie:workflow:0.5">
|
2020-03-18 15:43:54 +01:00
|
|
|
|
2019-10-28 12:12:50 +01:00
|
|
|
<parameters>
|
|
|
|
<property>
|
|
|
|
<name>sourcePath</name>
|
|
|
|
<description>the source path</description>
|
|
|
|
</property>
|
2019-11-07 17:38:09 +01:00
|
|
|
<property>
|
|
|
|
<name>hive_db_name</name>
|
|
|
|
<description>the target hive database name</description>
|
|
|
|
</property>
|
2019-10-29 18:10:20 +01:00
|
|
|
<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>
|
2020-04-21 16:12:20 +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>
|
|
|
|
<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>
|
2019-10-28 12:12:50 +01:00
|
|
|
</parameters>
|
|
|
|
|
2020-03-18 15:43:54 +01:00
|
|
|
<global>
|
|
|
|
<job-tracker>${jobTracker}</job-tracker>
|
|
|
|
<name-node>${nameNode}</name-node>
|
|
|
|
<configuration>
|
|
|
|
<property>
|
|
|
|
<name>mapreduce.job.queuename</name>
|
|
|
|
<value>${queueName}</value>
|
|
|
|
</property>
|
|
|
|
<property>
|
|
|
|
<name>oozie.launcher.mapred.job.queue.name</name>
|
|
|
|
<value>${oozieLauncherQueueName}</value>
|
|
|
|
</property>
|
2020-04-21 16:12:20 +02:00
|
|
|
<property>
|
|
|
|
<name>oozie.action.sharelib.for.spark</name>
|
|
|
|
<value>${oozieActionShareLibForSpark2}</value>
|
|
|
|
</property>
|
2020-03-18 15:43:54 +01:00
|
|
|
</configuration>
|
|
|
|
</global>
|
|
|
|
|
2020-03-23 17:40:39 +01:00
|
|
|
<start to="MapGraphAsHiveDB"/>
|
2019-10-28 12:12:50 +01:00
|
|
|
|
|
|
|
<kill name="Kill">
|
|
|
|
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
|
|
</kill>
|
|
|
|
|
2020-03-23 17:40:39 +01:00
|
|
|
<action name="MapGraphAsHiveDB">
|
2019-10-28 12:12:50 +01:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
2020-03-18 15:43:54 +01:00
|
|
|
<master>yarn</master>
|
2019-10-28 12:12:50 +01:00
|
|
|
<mode>cluster</mode>
|
2020-03-23 17:40:39 +01:00
|
|
|
<name>MapGraphAsHiveDB</name>
|
2020-04-08 13:24:43 +02:00
|
|
|
<class>eu.dnetlib.dhp.oa.graph.GraphHiveImporterJob</class>
|
2019-10-29 18:10:20 +01:00
|
|
|
<jar>dhp-graph-mapper-${projectVersion}.jar</jar>
|
2020-03-04 10:56:50 +01:00
|
|
|
<spark-opts>
|
2020-04-21 16:12:20 +02:00
|
|
|
--executor-memory=${sparkExecutorMemory}
|
|
|
|
--executor-cores=${sparkExecutorCores}
|
2020-03-04 10:56:50 +01:00
|
|
|
--driver-memory=${sparkDriverMemory}
|
2020-04-21 16:12:20 +02:00
|
|
|
--conf spark.extraListeners=${spark2ExtraListeners}
|
|
|
|
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
|
|
|
|
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
|
|
|
|
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
|
|
|
|
--conf spark.sql.warehouse.dir=${sparkSqlWarehouseDir}
|
2020-03-13 15:53:56 +01:00
|
|
|
</spark-opts>
|
2020-03-26 19:44:19 +01:00
|
|
|
<arg>--sourcePath</arg><arg>${sourcePath}</arg>
|
|
|
|
<arg>--hive_db_name</arg><arg>${hive_db_name}</arg>
|
|
|
|
<arg>--hive_metastore_uris</arg><arg>${hive_metastore_uris}</arg>
|
2019-10-28 12:12:50 +01:00
|
|
|
</spark>
|
2020-03-04 10:56:50 +01:00
|
|
|
<ok to="PostProcessing"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="PostProcessing">
|
2020-03-23 11:14:54 +01:00
|
|
|
<hive2 xmlns="uri:oozie:hive2-action:0.1">
|
|
|
|
<job-tracker>${jobTracker}</job-tracker>
|
|
|
|
<name-node>${nameNode}</name-node>
|
2020-03-18 15:43:54 +01:00
|
|
|
<configuration>
|
|
|
|
<property>
|
2020-03-23 17:40:39 +01:00
|
|
|
<name>hive.metastore.uris</name>
|
|
|
|
<value>${hive_metastore_uris}</value>
|
2020-03-18 15:43:54 +01:00
|
|
|
</property>
|
|
|
|
</configuration>
|
2020-03-23 17:40:39 +01:00
|
|
|
<jdbc-url>${hive_jdbc_url}/${hive_db_name}</jdbc-url>
|
2020-03-18 15:43:54 +01:00
|
|
|
<script>lib/scripts/postprocessing.sql</script>
|
2020-03-04 10:56:50 +01:00
|
|
|
<param>hive_db_name=${hive_db_name}</param>
|
2020-03-23 11:14:54 +01:00
|
|
|
</hive2>
|
2019-10-28 12:12:50 +01:00
|
|
|
<ok to="End"/>
|
2019-10-29 18:10:20 +01:00
|
|
|
<error to="Kill"/>
|
2019-10-28 12:12:50 +01:00
|
|
|
</action>
|
|
|
|
|
|
|
|
<end name="End"/>
|
|
|
|
</workflow-app>
|