Merge pull request 'Use SparkSQL in place of Hive for executing step16-createIndicatorsTables.sql of stats update wf' (#386) from stats_with_spark_sql into beta

Reviewed-on: #386
This commit is contained in:
Claudio Atzori 2024-01-29 09:11:59 +01:00
commit f804c58bc7
3 changed files with 436 additions and 395 deletions

View File

@ -8,6 +8,11 @@
<modelVersion>4.0.0</modelVersion>
<artifactId>dhp-stats-update</artifactId>
<dependencies>
<dependency>
<groupId>eu.dnetlib.dhp</groupId>
<artifactId>dhp-common</artifactId>
<version>${project.version}</version>
</dependency>
<dependency>
<groupId>org.apache.spark</groupId>
<artifactId>spark-core_${scala.binary.version}</artifactId>

View File

@ -64,6 +64,26 @@
<name>hadoop_user_name</name>
<description>user name of the wf owner</description>
</property>
<property>
<name>sparkSqlWarehouseDir</name>
</property>
<!-- General oozie workflow properties -->
<property>
<name>sparkClusterOpts</name>
<value>--conf spark.network.timeout=600 --conf spark.extraListeners= --conf spark.sql.queryExecutionListeners= --conf spark.yarn.historyServer.address=http://iis-cdh5-test-m3.ocean.icm.edu.pl:18088 --conf spark.eventLog.dir=hdfs://nameservice1/user/spark/applicationHistory</value>
<description>spark cluster-wide options</description>
</property>
<property>
<name>sparkResourceOpts</name>
<value>--executor-memory=6G --conf spark.executor.memoryOverhead=4G --executor-cores=6 --driver-memory=8G --driver-cores=4</value>
<description>spark resource options</description>
</property>
<property>
<name>sparkApplicationOpts</name>
<value>--conf spark.sql.shuffle.partitions=3840</value>
<description>spark resource options</description>
</property>
</parameters>
<global>
@ -75,17 +95,21 @@
<value>${hive_metastore_uris}</value>
</property>
<property>
<name>hive.txn.timeout</name>
<value>${hive_timeout}</value>
<name>hive.txn.timeout</name>
<value>${hive_timeout}</value>
</property>
<property>
<name>hive.mapjoin.followby.gby.localtask.max.memory.usage</name>
<value>0.80</value>
</property>
<property>
<name>mapred.job.queue.name</name>
<value>analytics</value>
</property>
<property>
<name>oozie.action.sharelib.for.spark</name>
<value>${oozieActionShareLibForSpark2}</value>
</property>
<property>
<name>mapred.job.queue.name</name>
<value>analytics</value>
</property>
</configuration>
</global>
@ -133,164 +157,164 @@
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<jdbc-url>${hive_jdbc_url}</jdbc-url>
<script>scripts/step1.sql</script>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
</hive2>
<ok to="Step2"/>
<error to="Kill"/>
</action>
<action name="Step2">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<jdbc-url>${hive_jdbc_url}</jdbc-url>
<script>scripts/step2.sql</script>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
</hive2>
<ok to="Step3"/>
<error to="Kill"/>
</action>
<action name="Step3">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<jdbc-url>${hive_jdbc_url}</jdbc-url>
<script>scripts/step3.sql</script>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
</hive2>
<ok to="Step4"/>
<error to="Kill"/>
</action>
<action name="Step4">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<jdbc-url>${hive_jdbc_url}</jdbc-url>
<script>scripts/step4.sql</script>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
</hive2>
<ok to="Step5"/>
<error to="Kill"/>
</action>
<action name="Step5">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<jdbc-url>${hive_jdbc_url}</jdbc-url>
<script>scripts/step5.sql</script>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
</hive2>
<ok to="Step6"/>
<error to="Kill"/>
</action>
<action name="Step6">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<jdbc-url>${hive_jdbc_url}</jdbc-url>
<script>scripts/step6.sql</script>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
</hive2>
<ok to="Step7"/>
<error to="Kill"/>
</action>
<action name="Step7">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<jdbc-url>${hive_jdbc_url}</jdbc-url>
<script>scripts/step7.sql</script>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
</hive2>
<ok to="Step8"/>
<error to="Kill"/>
</action>
<action name="Step8">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<jdbc-url>${hive_jdbc_url}</jdbc-url>
<script>scripts/step8.sql</script>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
</hive2>
<ok to="Step9"/>
<error to="Kill"/>
</action>
<action name="Step9">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<jdbc-url>${hive_jdbc_url}</jdbc-url>
<script>scripts/step9.sql</script>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
</hive2>
<ok to="Step10"/>
<error to="Kill"/>
</action>
<action name="Step10">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<jdbc-url>${hive_jdbc_url}</jdbc-url>
<script>scripts/step10.sql</script>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
<param>external_stats_db_name=${external_stats_db_name}</param>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
<param>external_stats_db_name=${external_stats_db_name}</param>
</hive2>
<ok to="Step11"/>
<error to="Kill"/>
</action>
</action>
<action name="Step11">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<jdbc-url>${hive_jdbc_url}</jdbc-url>
<script>scripts/step11.sql</script>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
<param>external_stats_db_name=${external_stats_db_name}</param>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
<param>external_stats_db_name=${external_stats_db_name}</param>
</hive2>
<ok to="Step12"/>
<error to="Kill"/>
</action>
</action>
<action name="Step12">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<jdbc-url>${hive_jdbc_url}</jdbc-url>
<script>scripts/step12.sql</script>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
</hive2>
<ok to="Step13"/>
<error to="Kill"/>
</action>
<action name="Step13">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<jdbc-url>${hive_jdbc_url}</jdbc-url>
<script>scripts/step13.sql</script>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
</hive2>
<ok to="Step14"/>
<error to="Kill"/>
</action>
<action name="Step14">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<jdbc-url>${hive_jdbc_url}</jdbc-url>
<script>scripts/step14.sql</script>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
</hive2>
<ok to="Step15"/>
<error to="Kill"/>
</action>
<action name="Step15">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<jdbc-url>${hive_jdbc_url}</jdbc-url>
<script>scripts/step15.sql</script>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
<param>stats_db_name=${stats_db_name}</param>
<param>openaire_db_name=${openaire_db_name}</param>
</hive2>
<ok to="Step15_5"/>
<error to="Kill"/>
@ -322,12 +346,23 @@
</action>
<action name="Step16-createIndicatorsTables">
<hive2 xmlns="uri:oozie:hive2-action:0.1">
<jdbc-url>${hive_jdbc_url}</jdbc-url>
<script>scripts/step16-createIndicatorsTables.sql</script>
<param>stats_db_name=${stats_db_name}</param>
<param>external_stats_db_name=${external_stats_db_name}</param>
</hive2>
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn</master>
<mode>cluster</mode>
<name>Step16-createIndicatorsTables</name>
<class>eu.dnetlib.dhp.oozie.RunSQLSparkJob</class>
<jar>dhp-stats-update-${projectVersion}.jar</jar>
<spark-opts>
--conf spark.sql.warehouse.dir=${sparkSqlWarehouseDir}
${sparkClusterOpts}
${sparkResourceOpts}
${sparkApplicationOpts}
</spark-opts>
<arg>--hiveMetastoreUris</arg><arg>${hive_metastore_uris}</arg>
<arg>--sql</arg><arg>eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16-createIndicatorsTables.sql</arg>
<arg>--stats_db_name</arg><arg>${stats_db_name}</arg>
<arg>--external_stats_db_name</arg><arg>${external_stats_db_name}</arg>
</spark>
<ok to="Step16_1-definitions"/>
<error to="Kill"/>
</action>
@ -387,18 +422,18 @@
<error to="Kill"/>
</action>
<!-- <action name="step20-createMonitorDB-post">-->
<!-- <shell xmlns="uri:oozie:shell-action:0.1">-->
<!-- <job-tracker>${jobTracker}</job-tracker>-->
<!-- <name-node>${nameNode}</name-node>-->
<!-- <exec>monitor-post.sh</exec>-->
<!-- <argument>${monitor_db_name}</argument>-->
<!-- <argument>${monitor_db_shadow_name}</argument>-->
<!-- <file>monitor-post.sh</file>-->
<!-- </shell>-->
<!-- <ok to="step21-createObservatoryDB-pre"/>-->
<!-- <error to="Kill"/>-->
<!-- </action>-->
<!-- <action name="step20-createMonitorDB-post">-->
<!-- <shell xmlns="uri:oozie:shell-action:0.1">-->
<!-- <job-tracker>${jobTracker}</job-tracker>-->
<!-- <name-node>${nameNode}</name-node>-->
<!-- <exec>monitor-post.sh</exec>-->
<!-- <argument>${monitor_db_name}</argument>-->
<!-- <argument>${monitor_db_shadow_name}</argument>-->
<!-- <file>monitor-post.sh</file>-->
<!-- </shell>-->
<!-- <ok to="step21-createObservatoryDB-pre"/>-->
<!-- <error to="Kill"/>-->
<!-- </action>-->
<action name="step21-createObservatoryDB-pre">
<shell xmlns="uri:oozie:shell-action:0.1">
@ -443,8 +478,8 @@
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<exec>copyDataToImpalaCluster.sh</exec>
<!-- <env-var>HADOOP_USER_NAME=${wf:user()}</env-var>-->
<!-- <argument>${external_stats_db_name}</argument>-->
<!-- <env-var>HADOOP_USER_NAME=${wf:user()}</env-var>-->
<!-- <argument>${external_stats_db_name}</argument>-->
<argument>${stats_db_name}</argument>
<argument>${monitor_db_name}</argument>
<argument>${observatory_db_name}</argument>
@ -505,4 +540,4 @@
</action>
<end name="End"/>
</workflow-app>
</workflow-app>