From 73bd1938a5a8dfead4035f50fded1108cf45a281 Mon Sep 17 00:00:00 2001
From: Claudio Atzori <claudio.atzori@isti.cnr.it>
Date: Wed, 5 Jun 2024 12:17:35 +0200
Subject: [PATCH] [graph2hive] use sparkExecutorMemory to define also the
 memoryOverhead

---
 .../dhp/oa/graph/hive/oozie_app/workflow.xml  | 30 ++++++++++++++-----
 1 file changed, 23 insertions(+), 7 deletions(-)

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 4468382be..eec67fc5c 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
@@ -116,17 +116,19 @@
                 --executor-memory=${sparkExecutorMemory}
                 --executor-cores=${sparkExecutorCores}
                 --driver-memory=${sparkDriverMemory}
+                --conf spark.executor.memoryOverhead=${sparkExecutorMemory}
                 --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}
+                --conf spark.sql.shuffle.partitions=10000
             </spark-opts>
             <arg>--inputPath</arg><arg>${inputPath}/publication</arg>
             <arg>--hiveDbName</arg><arg>${hiveDbName}</arg>
             <arg>--className</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
             <arg>--hiveMetastoreUris</arg><arg>${hiveMetastoreUris}</arg>
-            <arg>--numPartitions</arg><arg>8000</arg>
+            <arg>--numPartitions</arg><arg>10000</arg>
         </spark>
         <ok to="join_import"/>
         <error to="Kill"/>
@@ -143,17 +145,19 @@
                 --executor-memory=${sparkExecutorMemory}
                 --executor-cores=${sparkExecutorCores}
                 --driver-memory=${sparkDriverMemory}
+                --conf spark.executor.memoryOverhead=${sparkExecutorMemory}
                 --conf spark.extraListeners=${spark2ExtraListeners}
                 --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                 --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                 --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                 --conf spark.sql.warehouse.dir=${sparkSqlWarehouseDir}
+                --conf spark.sql.shuffle.partitions=4000
             </spark-opts>
             <arg>--inputPath</arg><arg>${inputPath}/dataset</arg>
             <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>
+            <arg>--numPartitions</arg><arg>8000</arg>
         </spark>
         <ok to="join_import"/>
         <error to="Kill"/>
@@ -170,11 +174,13 @@
                 --executor-memory=${sparkExecutorMemory}
                 --executor-cores=${sparkExecutorCores}
                 --driver-memory=${sparkDriverMemory}
+                --conf spark.executor.memoryOverhead=${sparkExecutorMemory}
                 --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}
+                --conf spark.sql.shuffle.partitions=8000
             </spark-opts>
             <arg>--inputPath</arg><arg>${inputPath}/otherresearchproduct</arg>
             <arg>--hiveDbName</arg><arg>${hiveDbName}</arg>
@@ -197,17 +203,19 @@
                 --executor-memory=${sparkExecutorMemory}
                 --executor-cores=${sparkExecutorCores}
                 --driver-memory=${sparkDriverMemory}
+                --conf spark.executor.memoryOverhead=${sparkExecutorMemory}
                 --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}
+                --conf spark.sql.shuffle.partitions=1000
             </spark-opts>
             <arg>--inputPath</arg><arg>${inputPath}/software</arg>
             <arg>--hiveDbName</arg><arg>${hiveDbName}</arg>
             <arg>--className</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
             <arg>--hiveMetastoreUris</arg><arg>${hiveMetastoreUris}</arg>
-            <arg>--numPartitions</arg><arg>300</arg>
+            <arg>--numPartitions</arg><arg>1000</arg>
         </spark>
         <ok to="join_import"/>
         <error to="Kill"/>
@@ -224,17 +232,19 @@
                 --executor-memory=${sparkExecutorMemory}
                 --executor-cores=${sparkExecutorCores}
                 --driver-memory=${sparkDriverMemory}
+                --conf spark.executor.memoryOverhead=${sparkExecutorMemory}
                 --conf spark.extraListeners=${spark2ExtraListeners}
                 --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                 --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                 --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                 --conf spark.sql.warehouse.dir=${sparkSqlWarehouseDir}
+                --conf spark.sql.shuffle.partitions=200
             </spark-opts>
             <arg>--inputPath</arg><arg>${inputPath}/datasource</arg>
             <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>
+            <arg>--numPartitions</arg><arg>200</arg>
         </spark>
         <ok to="join_import"/>
         <error to="Kill"/>
@@ -251,17 +261,19 @@
                 --executor-memory=${sparkExecutorMemory}
                 --executor-cores=${sparkExecutorCores}
                 --driver-memory=${sparkDriverMemory}
+                --conf spark.executor.memoryOverhead=${sparkExecutorMemory}
                 --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}
+                --conf spark.sql.shuffle.partitions=1000
             </spark-opts>
             <arg>--inputPath</arg><arg>${inputPath}/organization</arg>
             <arg>--hiveDbName</arg><arg>${hiveDbName}</arg>
             <arg>--className</arg><arg>eu.dnetlib.dhp.schema.oaf.Organization</arg>
             <arg>--hiveMetastoreUris</arg><arg>${hiveMetastoreUris}</arg>
-            <arg>--numPartitions</arg><arg>400</arg>
+            <arg>--numPartitions</arg><arg>1000</arg>
         </spark>
         <ok to="join_import"/>
         <error to="Kill"/>
@@ -278,17 +290,19 @@
                 --executor-memory=${sparkExecutorMemory}
                 --executor-cores=${sparkExecutorCores}
                 --driver-memory=${sparkDriverMemory}
+                --conf spark.executor.memoryOverhead=${sparkExecutorMemory}
                 --conf spark.extraListeners=${spark2ExtraListeners}
                 --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
                 --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
                 --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
                 --conf spark.sql.warehouse.dir=${sparkSqlWarehouseDir}
+                --conf spark.sql.shuffle.partitions=1000
             </spark-opts>
             <arg>--inputPath</arg><arg>${inputPath}/project</arg>
             <arg>--hiveDbName</arg><arg>${hiveDbName}</arg>
             <arg>--className</arg><arg>eu.dnetlib.dhp.schema.oaf.Project</arg>
             <arg>--hiveMetastoreUris</arg><arg>${hiveMetastoreUris}</arg>
-            <arg>--numPartitions</arg><arg>100</arg>
+            <arg>--numPartitions</arg><arg>1000</arg>
         </spark>
         <ok to="join_import"/>
         <error to="Kill"/>
@@ -305,17 +319,19 @@
                 --executor-memory=${sparkExecutorMemory}
                 --executor-cores=${sparkExecutorCores}
                 --driver-memory=${sparkDriverMemory}
+                --conf spark.executor.memoryOverhead=${sparkExecutorMemory}
                 --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}
+                --conf spark.sql.shuffle.partitions=15000
             </spark-opts>
             <arg>--inputPath</arg><arg>${inputPath}/relation</arg>
             <arg>--hiveDbName</arg><arg>${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>
+            <arg>--numPartitions</arg><arg>15000</arg>
         </spark>
         <ok to="join_import"/>
         <error to="Kill"/>