-

dhp-workflows/dhp-doiboost/src/main/resources/eu/dnetlib/dhp/doiboost/crossref_dump_reader/oozie_app/workflow.xml

@@ -24,7 +24,7 @@
 
     </parameters>
 
-    <start to="generateCrossrefDataset"/>
+    <start to="GenerateCrossrefDataset"/>
 
     <kill name="Kill">
         <message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
@@ -40,16 +40,16 @@
             <arg>--workingPath</arg><arg>${crossrefDumpPath}</arg>
             <arg>--outputPath</arg><arg>${workingDir}/files/</arg>
         </java>
-        <ok to="generateCrossrefDataset"/>
+        <ok to="UnpackCrossrefEntries"/>
         <error to="Kill"/>
     </action>
 
-    <action name="generateCrossrefDataset">
+    <action name="UnpackCrossrefEntries">
         <spark xmlns="uri:oozie:spark-action:0.2">
             <master>yarn-cluster</master>
             <mode>cluster</mode>
             <name>SparkGenerateCrossrefDataset</name>
-            <class>eu.dnetlib.doiboost.crossref.GenerateCrossrefDataset</class>
+            <class>eu.dnetlib.doiboost.crossref.UnpackCrtossrefEntries</class>
             <jar>dhp-doiboost-${projectVersion}.jar</jar>
             <spark-opts>
                 --executor-memory=${sparkExecutorMemory}
@@ -62,9 +62,35 @@
                 --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
             </spark-opts>
             <arg>--master</arg><arg>yarn-cluster</arg>
-            <arg>--sourcePath</arg><arg>${workingDir}/files</arg>
+            <arg>--sourcePath</arg><arg>${crossrefDumpPath}/files</arg>
             <arg>--targetPath</arg><arg>${inputPathCrossref}/crossref_ds</arg>
 
+        </spark>
+        <ok to="GenerateCrossrefDataset"/>
+        <error to="Kill"/>
+    </action>
+
+    <action name="GenerateCrossrefDataset">
+        <spark xmlns="uri:oozie:spark-action:0.2">
+            <master>yarn-cluster</master>
+            <mode>cluster</mode>
+            <name>SparkGenerateCrossrefDataset</name>
+            <class>eu.dnetlib.doiboost.crossref.GenerateCrossrefDataset</class>
+            <jar>dhp-doiboost-${projectVersion}.jar</jar>
+            <spark-opts>
+                --executor-memory=${sparkExecutorMemory}
+                --executor-cores=${sparkExecutorCores}
+                --driver-memory=${sparkDriverMemory}
+                --conf spark.sql.shuffle.partitions=3840
+                --conf spark.extraListeners=${spark2ExtraListeners}
+                --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
+                --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
+                --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
+            </spark-opts>
+            <arg>--master</arg><arg>yarn-cluster</arg>
+            <arg>--sourcePath</arg><arg>${inputPathCrossref}/crossref_ds</arg>
+            <arg>--targetPath</arg><arg>${inputPathCrossref}/crossref_ds_updates</arg>
+
         </spark>
         <ok to="End"/>
         <error to="Kill"/>