From cfd753217cac987f85841baad42064b4047c71e5 Mon Sep 17 00:00:00 2001
From: Claudio Atzori <claudio.atzori@isti.cnr.it>
Date: Wed, 27 May 2020 12:44:01 +0200
Subject: [PATCH] repartition the join_entities in 24k files

---
 .../eu/dnetlib/dhp/oa/provision/oozie_app/workflow.xml        | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/dhp-workflows/dhp-graph-provision/src/main/resources/eu/dnetlib/dhp/oa/provision/oozie_app/workflow.xml b/dhp-workflows/dhp-graph-provision/src/main/resources/eu/dnetlib/dhp/oa/provision/oozie_app/workflow.xml
index 6983ecf53..5d1889439 100644
--- a/dhp-workflows/dhp-graph-provision/src/main/resources/eu/dnetlib/dhp/oa/provision/oozie_app/workflow.xml
+++ b/dhp-workflows/dhp-graph-provision/src/main/resources/eu/dnetlib/dhp/oa/provision/oozie_app/workflow.xml
@@ -362,7 +362,7 @@
             <arg>--inputGraphRootPath</arg><arg>${inputGraphRootPath}</arg>
             <arg>--inputRelatedEntitiesPath</arg><arg>${workingDir}/join_partial</arg>
             <arg>--outputPath</arg><arg>${workingDir}/join_entities</arg>
-            <arg>--numPartitions</arg><arg>12000</arg>
+            <arg>--numPartitions</arg><arg>24000</arg>
         </spark>
         <ok to="adjancency_lists"/>
         <error to="Kill"/>
@@ -386,7 +386,7 @@
                 --conf spark.sql.shuffle.partitions=7680
                 --conf spark.network.timeout=${sparkNetworkTimeout}
             </spark-opts>
-            <arg>--inputPath</arg> <arg>${workingDir}/join_entities</arg>
+            <arg>--inputPath</arg><arg>${workingDir}/join_entities</arg>
             <arg>--outputPath</arg><arg>${workingDir}/joined</arg>
         </spark>
         <ok to="convert_to_xml"/>