From 3585e53da3818289db8e783f92642fa9520ae1a7 Mon Sep 17 00:00:00 2001
From: "miriam.baglioni" <miriam.baglioni@isti.cnr.it>
Date: Fri, 18 Jun 2021 09:41:23 +0200
Subject: [PATCH] changed to split in two steps the generation of the crossref
 dataset

---
 .../dhp/doiboost/oozie_app/workflow.xml       | 33 +++++++++++++++++--
 1 file changed, 30 insertions(+), 3 deletions(-)

diff --git a/dhp-workflows/dhp-doiboost/src/main/resources/eu/dnetlib/dhp/doiboost/oozie_app/workflow.xml b/dhp-workflows/dhp-doiboost/src/main/resources/eu/dnetlib/dhp/doiboost/oozie_app/workflow.xml
index 5a943c673d..fa47e142d0 100644
--- a/dhp-workflows/dhp-doiboost/src/main/resources/eu/dnetlib/dhp/doiboost/oozie_app/workflow.xml
+++ b/dhp-workflows/dhp-doiboost/src/main/resources/eu/dnetlib/dhp/doiboost/oozie_app/workflow.xml
@@ -142,6 +142,32 @@
             <arg>--workingPath</arg><arg>${crossrefDumpPath}</arg>
             <arg>--outputPath</arg><arg>${crossrefDumpPath}/files/</arg>
         </java>
+        <ok to="UnpackCrossrefEntries"/>
+        <error to="Kill"/>
+    </action>
+
+    <action name="UnpackCrossrefEntries">
+        <spark xmlns="uri:oozie:spark-action:0.2">
+            <master>yarn-cluster</master>
+            <mode>cluster</mode>
+            <name>SparkUnpackCrossrefEntries</name>
+            <class>eu.dnetlib.doiboost.crossref.UnpackCrtossrefEntries</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}${spark2EventLogDir}
+            </spark-opts>
+            <arg>--master</arg><arg>yarn-cluster</arg>
+            <arg>--sourcePath</arg><arg>${crossrefDumpPath}/files</arg>
+            <arg>--targetPath</arg><arg>${crossrefDumpPath}/crossref_unpack/</arg>
+
+        </spark>
         <ok to="GenerateCrossrefDataset"/>
         <error to="Kill"/>
     </action>
@@ -155,7 +181,7 @@
             <jar>dhp-doiboost-${projectVersion}.jar</jar>
             <spark-opts>
                 --executor-memory=7G
-                --executor-cores=4
+                --executor-cores=2
                 --driver-memory=7G
                 --conf spark.sql.shuffle.partitions=3840
                 --conf spark.extraListeners=${spark2ExtraListeners}
@@ -164,7 +190,7 @@
                 --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
             </spark-opts>
             <arg>--master</arg><arg>yarn-cluster</arg>
-            <arg>--sourcePath</arg><arg>${crossrefDumpPath}/files/</arg>
+            <arg>--sourcePath</arg><arg>${crossrefDumpPath}/crossref_unpack/</arg>
             <arg>--targetPath</arg><arg>${inputPathCrossref}/crossref_ds</arg>
 
         </spark>
@@ -174,7 +200,8 @@
 
         <action name="removeFiles">
             <fs>
-                <delete path="${crossrefDumpPath}/files"/>
+<!--                <delete path="${crossrefDumpPath}/files"/>-->
+                    <delete path="${crossrefDumpPath}/crossref_unpack/"/>
             </fs>
             <ok to="ResetMagWorkingPath"/>
             <error to="Kill"/>