From ae7cd0735a8528cd419ce2863077d436bd34cd0d Mon Sep 17 00:00:00 2001
From: Claudio Atzori <claudio.atzori@isti.cnr.it>
Date: Fri, 14 Oct 2022 15:47:58 +0200
Subject: [PATCH] [graph2hive] more partitions

---
 .../eu/dnetlib/dhp/oa/graph/hive/oozie_app/workflow.xml    | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/oa/graph/hive/oozie_app/workflow.xml b/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/oa/graph/hive/oozie_app/workflow.xml
index ba5f4f375..4468382be 100644
--- a/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/oa/graph/hive/oozie_app/workflow.xml
+++ b/dhp-workflows/dhp-graph-mapper/src/main/resources/eu/dnetlib/dhp/oa/graph/hive/oozie_app/workflow.xml
@@ -126,6 +126,7 @@
             <arg>--hiveDbName</arg><arg>${hiveDbName}</arg>
             <arg>--className</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
             <arg>--hiveMetastoreUris</arg><arg>${hiveMetastoreUris}</arg>
+            <arg>--numPartitions</arg><arg>8000</arg>
         </spark>
         <ok to="join_import"/>
         <error to="Kill"/>
@@ -152,6 +153,7 @@
             <arg>--hiveDbName</arg><arg>${hiveDbName}</arg>
             <arg>--className</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
             <arg>--hiveMetastoreUris</arg><arg>${hiveMetastoreUris}</arg>
+            <arg>--numPartitions</arg><arg>4000</arg>
         </spark>
         <ok to="join_import"/>
         <error to="Kill"/>
@@ -178,6 +180,7 @@
             <arg>--hiveDbName</arg><arg>${hiveDbName}</arg>
             <arg>--className</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
             <arg>--hiveMetastoreUris</arg><arg>${hiveMetastoreUris}</arg>
+            <arg>--numPartitions</arg><arg>3000</arg>
         </spark>
         <ok to="join_import"/>
         <error to="Kill"/>
@@ -204,6 +207,7 @@
             <arg>--hiveDbName</arg><arg>${hiveDbName}</arg>
             <arg>--className</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
             <arg>--hiveMetastoreUris</arg><arg>${hiveMetastoreUris}</arg>
+            <arg>--numPartitions</arg><arg>300</arg>
         </spark>
         <ok to="join_import"/>
         <error to="Kill"/>
@@ -230,6 +234,7 @@
             <arg>--hiveDbName</arg><arg>${hiveDbName}</arg>
             <arg>--className</arg><arg>eu.dnetlib.dhp.schema.oaf.Datasource</arg>
             <arg>--hiveMetastoreUris</arg><arg>${hiveMetastoreUris}</arg>
+            <arg>--numPartitions</arg><arg>100</arg>
         </spark>
         <ok to="join_import"/>
         <error to="Kill"/>
@@ -256,6 +261,7 @@
             <arg>--hiveDbName</arg><arg>${hiveDbName}</arg>
             <arg>--className</arg><arg>eu.dnetlib.dhp.schema.oaf.Organization</arg>
             <arg>--hiveMetastoreUris</arg><arg>${hiveMetastoreUris}</arg>
+            <arg>--numPartitions</arg><arg>400</arg>
         </spark>
         <ok to="join_import"/>
         <error to="Kill"/>
@@ -309,6 +315,7 @@
             <arg>--hiveDbName</arg><arg>${hiveDbName}</arg>
             <arg>--className</arg><arg>eu.dnetlib.dhp.schema.oaf.Relation</arg>
             <arg>--hiveMetastoreUris</arg><arg>${hiveMetastoreUris}</arg>
+            <arg>--numPartitions</arg><arg>10000</arg>
         </spark>
         <ok to="join_import"/>
         <error to="Kill"/>