sabeel
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Pull Requests Activity

[broker] fine tuned the workflow memory settings

This commit is contained in:
Claudio Atzori 2024-07-12 10:27:24 +02:00
parent 5ab409dcab
commit 2f13683285
1 changed files with 31 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

45
dhp-workflows/dhp-broker-events/src/main/resources/eu/dnetlib/dhp/broker/oa/generate_all/oozie_app/workflow.xml
View File

@ -7,7 +7,7 @@
</property> </property>
<property> <property>
<name>outputDir</name> <name>outputDir</name>
<description>the path where the the generated data will be stored</description> <description>the path where the generated data will be stored</description>
</property> </property>
<property> <property>
<name>datasourceIdWhitelist</name> <name>datasourceIdWhitelist</name>
@ -185,11 +185,12 @@
--executor-cores=${sparkExecutorCores} --executor-cores=${sparkExecutorCores}
--executor-memory=${sparkExecutorMemory} --executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory} --driver-memory=${sparkDriverMemory}
--conf spark.executor.memoryOverhead=${sparkExecutorMemory}
--conf spark.extraListeners=${spark2ExtraListeners} --conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=3840 --conf spark.sql.shuffle.partitions=5000
</spark-opts> </spark-opts>
<arg>--graphPath</arg><arg>${graphInputPath}</arg> <arg>--graphPath</arg><arg>${graphInputPath}</arg>
<arg>--workingDir</arg><arg>${workingDir}</arg> <arg>--workingDir</arg><arg>${workingDir}</arg>
@ -209,11 +210,12 @@
--executor-cores=${sparkExecutorCores} --executor-cores=${sparkExecutorCores}
--executor-memory=${sparkExecutorMemory} --executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory} --driver-memory=${sparkDriverMemory}
--conf spark.executor.memoryOverhead=${sparkExecutorMemory}
--conf spark.extraListeners=${spark2ExtraListeners} --conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=3840 --conf spark.sql.shuffle.partitions=8000
</spark-opts> </spark-opts>
<arg>--graphPath</arg><arg>${graphInputPath}</arg> <arg>--graphPath</arg><arg>${graphInputPath}</arg>
<arg>--workingDir</arg><arg>${workingDir}</arg> <arg>--workingDir</arg><arg>${workingDir}</arg>
@ -234,11 +236,12 @@
--executor-cores=${sparkExecutorCores} --executor-cores=${sparkExecutorCores}
--executor-memory=${sparkExecutorMemory} --executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory} --driver-memory=${sparkDriverMemory}
--conf spark.executor.memoryOverhead=${sparkExecutorMemory}
--conf spark.extraListeners=${spark2ExtraListeners} --conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=3840 --conf spark.sql.shuffle.partitions=8000
</spark-opts> </spark-opts>
<arg>--graphPath</arg><arg>${graphInputPath}</arg> <arg>--graphPath</arg><arg>${graphInputPath}</arg>
<arg>--workingDir</arg><arg>${workingDir}</arg> <arg>--workingDir</arg><arg>${workingDir}</arg>
@ -258,11 +261,12 @@
--executor-cores=${sparkExecutorCores} --executor-cores=${sparkExecutorCores}
--executor-memory=${sparkExecutorMemory} --executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory} --driver-memory=${sparkDriverMemory}
--conf spark.executor.memoryOverhead=${sparkExecutorMemory}
--conf spark.extraListeners=${spark2ExtraListeners} --conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=3840 --conf spark.sql.shuffle.partitions=5000
</spark-opts> </spark-opts>
<arg>--graphPath</arg><arg>${graphInputPath}</arg> <arg>--graphPath</arg><arg>${graphInputPath}</arg>
<arg>--workingDir</arg><arg>${workingDir}</arg> <arg>--workingDir</arg><arg>${workingDir}</arg>
@ -282,11 +286,12 @@
--executor-cores=${sparkExecutorCores} --executor-cores=${sparkExecutorCores}
--executor-memory=${sparkExecutorMemory} --executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory} --driver-memory=${sparkDriverMemory}
--conf spark.executor.memoryOverhead=${sparkExecutorMemory}
--conf spark.extraListeners=${spark2ExtraListeners} --conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=3840 --conf spark.sql.shuffle.partitions=10000
</spark-opts> </spark-opts>
<arg>--graphPath</arg><arg>${graphInputPath}</arg> <arg>--graphPath</arg><arg>${graphInputPath}</arg>
<arg>--workingDir</arg><arg>${workingDir}</arg> <arg>--workingDir</arg><arg>${workingDir}</arg>
@ -306,11 +311,12 @@
--executor-cores=${sparkExecutorCores} --executor-cores=${sparkExecutorCores}
--executor-memory=${sparkExecutorMemory} --executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory} --driver-memory=${sparkDriverMemory}
--conf spark.executor.memoryOverhead=${sparkExecutorMemory}
--conf spark.extraListeners=${spark2ExtraListeners} --conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=3840 --conf spark.sql.shuffle.partitions=2000
</spark-opts> </spark-opts>
<arg>--graphPath</arg><arg>${graphInputPath}</arg> <arg>--graphPath</arg><arg>${graphInputPath}</arg>
<arg>--workingDir</arg><arg>${workingDir}</arg> <arg>--workingDir</arg><arg>${workingDir}</arg>
@ -332,11 +338,12 @@
--executor-cores=${sparkExecutorCores} --executor-cores=${sparkExecutorCores}
--executor-memory=${sparkExecutorMemory} --executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory} --driver-memory=${sparkDriverMemory}
--conf spark.executor.memoryOverhead=${sparkExecutorMemory}
--conf spark.extraListeners=${spark2ExtraListeners} --conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=3840 --conf spark.sql.shuffle.partitions=8000
</spark-opts> </spark-opts>
<arg>--graphPath</arg><arg>${graphInputPath}</arg> <arg>--graphPath</arg><arg>${graphInputPath}</arg>
<arg>--workingDir</arg><arg>${workingDir}</arg> <arg>--workingDir</arg><arg>${workingDir}</arg>
@ -356,11 +363,12 @@
--executor-cores=${sparkExecutorCores} --executor-cores=${sparkExecutorCores}
--executor-memory=${sparkExecutorMemory} --executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory} --driver-memory=${sparkDriverMemory}
--conf spark.executor.memoryOverhead=${sparkExecutorMemory}
--conf spark.extraListeners=${spark2ExtraListeners} --conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=3840 --conf spark.sql.shuffle.partitions=8000
</spark-opts> </spark-opts>
<arg>--graphPath</arg><arg>${graphInputPath}</arg> <arg>--graphPath</arg><arg>${graphInputPath}</arg>
<arg>--workingDir</arg><arg>${workingDir}</arg> <arg>--workingDir</arg><arg>${workingDir}</arg>
@ -380,11 +388,12 @@
--executor-cores=${sparkExecutorCores} --executor-cores=${sparkExecutorCores}
--executor-memory=${sparkExecutorMemory} --executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory} --driver-memory=${sparkDriverMemory}
--conf spark.executor.memoryOverhead=${sparkExecutorMemory}
--conf spark.extraListeners=${spark2ExtraListeners} --conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=3840 --conf spark.sql.shuffle.partitions=8000
</spark-opts> </spark-opts>
<arg>--graphPath</arg><arg>${graphInputPath}</arg> <arg>--graphPath</arg><arg>${graphInputPath}</arg>
<arg>--workingDir</arg><arg>${workingDir}</arg> <arg>--workingDir</arg><arg>${workingDir}</arg>
@ -404,11 +413,12 @@
--executor-cores=${sparkExecutorCores} --executor-cores=${sparkExecutorCores}
--executor-memory=${sparkExecutorMemory} --executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory} --driver-memory=${sparkDriverMemory}
--conf spark.executor.memoryOverhead=${sparkExecutorMemory}
--conf spark.extraListeners=${spark2ExtraListeners} --conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=3840 --conf spark.sql.shuffle.partitions=8000
</spark-opts> </spark-opts>
<arg>--graphPath</arg><arg>${graphInputPath}</arg> <arg>--graphPath</arg><arg>${graphInputPath}</arg>
<arg>--workingDir</arg><arg>${workingDir}</arg> <arg>--workingDir</arg><arg>${workingDir}</arg>
@ -428,11 +438,12 @@
--executor-cores=${sparkExecutorCores} --executor-cores=${sparkExecutorCores}
--executor-memory=${sparkExecutorMemory} --executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory} --driver-memory=${sparkDriverMemory}
--conf spark.executor.memoryOverhead=${sparkExecutorMemory}
--conf spark.extraListeners=${spark2ExtraListeners} --conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=3840 --conf spark.sql.shuffle.partitions=8000
</spark-opts> </spark-opts>
<arg>--graphPath</arg><arg>${graphInputPath}</arg> <arg>--graphPath</arg><arg>${graphInputPath}</arg>
<arg>--workingDir</arg><arg>${workingDir}</arg> <arg>--workingDir</arg><arg>${workingDir}</arg>
@ -452,11 +463,12 @@
--executor-cores=${sparkExecutorCores} --executor-cores=${sparkExecutorCores}
--executor-memory=${sparkExecutorMemory} --executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory} --driver-memory=${sparkDriverMemory}
--conf spark.executor.memoryOverhead=${sparkExecutorMemory}
--conf spark.extraListeners=${spark2ExtraListeners} --conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=3840 --conf spark.sql.shuffle.partitions=8000
</spark-opts> </spark-opts>
<arg>--graphPath</arg><arg>${graphInputPath}</arg> <arg>--graphPath</arg><arg>${graphInputPath}</arg>
<arg>--workingDir</arg><arg>${workingDir}</arg> <arg>--workingDir</arg><arg>${workingDir}</arg>
@ -476,11 +488,12 @@
--executor-cores=${sparkExecutorCores} --executor-cores=${sparkExecutorCores}
--executor-memory=${sparkExecutorMemory} --executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory} --driver-memory=${sparkDriverMemory}
--conf spark.executor.memoryOverhead=${sparkExecutorMemory}
--conf spark.extraListeners=${spark2ExtraListeners} --conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=3840 --conf spark.sql.shuffle.partitions=8000
</spark-opts> </spark-opts>
<arg>--workingDir</arg><arg>${workingDir}</arg> <arg>--workingDir</arg><arg>${workingDir}</arg>
<arg>--outputDir</arg><arg>${outputDir}</arg> <arg>--outputDir</arg><arg>${outputDir}</arg>
@ -503,6 +516,7 @@
<spark-opts> <spark-opts>
--executor-memory=${sparkExecutorMemory} --executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory} --driver-memory=${sparkDriverMemory}
--conf spark.executor.memoryOverhead=${sparkExecutorMemory}
--conf spark.dynamicAllocation.maxExecutors=${sparkMaxExecutorsForIndexing} --conf spark.dynamicAllocation.maxExecutors=${sparkMaxExecutorsForIndexing}
--conf spark.extraListeners=${spark2ExtraListeners} --conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
@ -535,6 +549,7 @@
--executor-cores=${sparkExecutorCores} --executor-cores=${sparkExecutorCores}
--executor-memory=${sparkExecutorMemory} --executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory} --driver-memory=${sparkDriverMemory}
--conf spark.executor.memoryOverhead=${sparkExecutorMemory}
--conf spark.extraListeners=${spark2ExtraListeners} --conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
@ -562,6 +577,7 @@
--executor-cores=${sparkExecutorCores} --executor-cores=${sparkExecutorCores}
--executor-memory=${sparkExecutorMemory} --executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory} --driver-memory=${sparkDriverMemory}
--conf spark.executor.memoryOverhead=${sparkExecutorMemory}
--conf spark.extraListeners=${spark2ExtraListeners} --conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress} --conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
@ -585,6 +601,7 @@
<spark-opts> <spark-opts>
--executor-memory=${sparkExecutorMemory} --executor-memory=${sparkExecutorMemory}
--driver-memory=${sparkDriverMemory} --driver-memory=${sparkDriverMemory}
--conf spark.executor.memoryOverhead=${sparkExecutorMemory}
--conf spark.dynamicAllocation.maxExecutors=${sparkMaxExecutorsForIndexing} --conf spark.dynamicAllocation.maxExecutors=${sparkMaxExecutorsForIndexing}
--conf spark.extraListeners=${spark2ExtraListeners} --conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners} --conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}