[Enrichment WF] first attempt to make it run in a single step

This commit is contained in:
Miriam Baglioni 2022-04-06 18:14:05 +02:00
parent e77d104951
commit 7406c88276
17 changed files with 242 additions and 196 deletions

View File

@ -111,6 +111,13 @@ public class SparkBulkTagJob {
.mode(SaveMode.Overwrite) .mode(SaveMode.Overwrite)
.option("compression", "gzip") .option("compression", "gzip")
.json(outputPath); .json(outputPath);
readPath(spark, outputPath, resultClazz)
.write()
.mode(SaveMode.Overwrite)
.option("compression","gzip")
.json(inputPath);
} }
public static <R> Dataset<R> readPath( public static <R> Dataset<R> readPath(

View File

@ -122,6 +122,12 @@ public class SparkResultToOrganizationFromIstRepoJob {
.mode(SaveMode.Append) .mode(SaveMode.Append)
.option("compression", "gzip") .option("compression", "gzip")
.json(outputPath); .json(outputPath);
readPath(spark, outputPath, Relation.class)
.write()
.mode(SaveMode.Overwrite)
.option("compression","gzip")
.json(inputPath.substring(0, inputPath.indexOf("/") + 1) + "relation");
} }
private static FlatMapFunction<Tuple2<KeyValueSet, KeyValueSet>, Relation> createRelationFn() { private static FlatMapFunction<Tuple2<KeyValueSet, KeyValueSet>, Relation> createRelationFn() {

View File

@ -0,0 +1,30 @@
<configuration>
<property>
<name>jobTracker</name>
<value>yarnRM</value>
</property>
<property>
<name>nameNode</name>
<value>hdfs://nameservice1</value>
</property>
<property>
<name>oozie.use.system.libpath</name>
<value>true</value>
</property>
<property>
<name>hiveMetastoreUris</name>
<value>thrift://iis-cdh5-test-m3.ocean.icm.edu.pl:9083</value>
</property>
<property>
<name>hiveJdbcUrl</name>
<value>jdbc:hive2://iis-cdh5-test-m3.ocean.icm.edu.pl:10000</value>
</property>
<property>
<name>hiveDbName</name>
<value>openaire</value>
</property>
<property>
<name>oozie.launcher.mapreduce.user.classpath.first</name>
<value>true</value>
</property>
</configuration>

View File

@ -0,0 +1,4 @@
## This is a classpath-based import file (this header is required)
orcid_propagation classpath eu/dnetlib/dhp/wf/subworkflows/orcidtoresultfromsemrel/oozie_app
dump_funder classpath eu/dnetlib/dhp/oa/graph/dump/wf/subworkflows/funder/oozie_app
dump_community classpath eu/dnetlib/dhp/oa/graph/dump/wf/subworkflows/community/oozie_app

View File

@ -0,0 +1,187 @@
<workflow-app name="dump_graph" xmlns="uri:oozie:workflow:0.5">
<parameters>
<property>
<name>sourcePath</name>
<description>the source path</description>
</property>
<property>
<name>allowedsemrels</name>
<description>the semantic relationships allowed for propagation</description>
</property>
<property>
<name>outputPath</name>
<description>the output path</description>
</property>
<property>
<name>isLookUpUrl</name>
<description>the isLookup service endpoint</description>
</property>
<property>
<name>pathMap</name>
<description>the json path associated to each selection field</description>
</property>
<property>
<name>hiveDbName</name>
<description>the target hive database name</description>
</property>
<property>
<name>hiveJdbcUrl</name>
<description>hive server jdbc url</description>
</property>
<property>
<name>hiveMetastoreUris</name>
<description>hive server metastore URIs</description>
</property>
<property>
<name>sparkDriverMemory</name>
<description>memory for driver process</description>
</property>
<property>
<name>sparkExecutorMemory</name>
<description>memory for individual executor</description>
</property>
<property>
<name>sparkExecutorCores</name>
<description>number of cores used by single executor</description>
</property>
<property>
<name>oozieActionShareLibForSpark2</name>
<description>oozie action sharelib for spark 2.*</description>
</property>
<property>
<name>spark2ExtraListeners</name>
<value>com.cloudera.spark.lineage.NavigatorAppListener</value>
<description>spark 2.* extra listeners classname</description>
</property>
<property>
<name>spark2SqlQueryExecutionListeners</name>
<value>com.cloudera.spark.lineage.NavigatorQueryListener</value>
<description>spark 2.* sql query execution listeners classname</description>
</property>
<property>
<name>spark2YarnHistoryServerAddress</name>
<description>spark 2.* yarn history server address</description>
</property>
<property>
<name>spark2EventLogDir</name>
<description>spark 2.* event log dir location</description>
</property>
</parameters>
<global>
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<configuration>
<property>
<name>mapreduce.job.queuename</name>
<value>${queueName}</value>
</property>
<property>
<name>oozie.launcher.mapred.job.queue.name</name>
<value>${oozieLauncherQueueName}</value>
</property>
<property>
<name>oozie.action.sharelib.for.spark</name>
<value>${oozieActionShareLibForSpark2}</value>
</property>
</configuration>
</global>
<start to="resumeFrom"/>
<kill name="Kill">
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill>
<decision name="resumeFrom">
<switch>
<case to="send_zenodo">${wf:conf('onlyUpload') eq true}</case>
<default to="orcid_propagation"/>
</switch>
</decision>
<action name="orcid_propagation">
<sub-workflow>
<app-path>${wf:appPath()}/orcid_propagation
</app-path>
<propagate-configuration/>
<configuration>
<property>
<name>sourcePath</name>
<value>${sourcePath}</value>
</property>
<property>
<name>allowedsemrels</name>
<value>${allowedsemrels}</value>
</property>
<property>
<name>outputPath</name>
<value>${outputPath}</value>
</property>
</configuration>
</sub-workflow>
<ok to="bulk_tagging" />
<error to="Kill" />
</action>
<action name="bulk_tagging">
<sub-workflow>
<app-path>${wf:appPath()}/bulk_tagging
</app-path>
<propagate-configuration/>
<configuration>
<property>
<name>sourcePath</name>
<value>${outputPath}</value>
</property>
<property>
<name>isLookUpUrl</name>
<value>${isLookUpUrl}</value>
</property>
<property>
<name>pathMap</name>
<value>${pathMap}</value>
</property>
<property>
<name>outputPath</name>
<value>${workingDir}/results</value>
</property>
</configuration>
</sub-workflow>
<ok to="affiliation_inst_repo" />
<error to="Kill" />
</action>
<action name="affiliation_inst_repo">
<sub-workflow>
<app-path>${wf:appPath()}/bulk_tagging
</app-path>
<propagate-configuration/>
<configuration>
<property>
<name>sourcePath</name>
<value>${outputPath}</value>
</property>
<property>
<name>outputPath</name>
<value>${workingDir}/relations</value>
</property>
<property>
<name>blacklist</name>
<value>${blacklist}</value>
</property>
</configuration>
</sub-workflow>
<ok to="bulk_tagging" />
<error to="Kill" />
</action>
<end name="End"/>
</workflow-app>

View File

@ -25,7 +25,7 @@
}, },
{ {
"paramName": "out", "paramName": "out",
"paramLongName": "outputPath", "paramLongName": "workingPath",
"paramDescription": "the path used to store temporary output files", "paramDescription": "the path used to store temporary output files",
"paramRequired": true "paramRequired": true
}, },

View File

@ -13,7 +13,7 @@
<description>the json path associated to each selection field</description> <description>the json path associated to each selection field</description>
</property> </property>
<property> <property>
<name>outputPath</name> <name>workingPath</name>
<description>the output path</description> <description>the output path</description>
</property> </property>
</parameters> </parameters>
@ -29,65 +29,12 @@
</configuration> </configuration>
</global> </global>
<start to="reset_outputpath"/> <start to="fork_exec_bulktag"/>
<kill name="Kill"> <kill name="Kill">
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message> <message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill> </kill>
<action name="reset_outputpath">
<fs>
<delete path="${outputPath}"/>
<mkdir path="${outputPath}"/>
</fs>
<ok to="copy_entities"/>
<error to="Kill"/>
</action>
<fork name="copy_entities">
<path start="copy_relation"/>
<path start="copy_organization"/>
<path start="copy_projects"/>
<path start="copy_datasources"/>
</fork>
<action name="copy_relation">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>${nameNode}/${sourcePath}/relation</arg>
<arg>${nameNode}/${outputPath}/relation</arg>
</distcp>
<ok to="copy_wait"/>
<error to="Kill"/>
</action>
<action name="copy_organization">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>${nameNode}/${sourcePath}/organization</arg>
<arg>${nameNode}/${outputPath}/organization</arg>
</distcp>
<ok to="copy_wait"/>
<error to="Kill"/>
</action>
<action name="copy_projects">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>${nameNode}/${sourcePath}/project</arg>
<arg>${nameNode}/${outputPath}/project</arg>
</distcp>
<ok to="copy_wait"/>
<error to="Kill"/>
</action>
<action name="copy_datasources">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>${nameNode}/${sourcePath}/datasource</arg>
<arg>${nameNode}/${outputPath}/datasource</arg>
</distcp>
<ok to="copy_wait"/>
<error to="Kill"/>
</action>
<join name="copy_wait" to="fork_exec_bulktag"/>
<fork name="fork_exec_bulktag"> <fork name="fork_exec_bulktag">
<path start="join_bulktag_publication"/> <path start="join_bulktag_publication"/>
@ -115,7 +62,7 @@
</spark-opts> </spark-opts>
<arg>--sourcePath</arg><arg>${sourcePath}/publication</arg> <arg>--sourcePath</arg><arg>${sourcePath}/publication</arg>
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg> <arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
<arg>--outputPath</arg><arg>${outputPath}/publication</arg> <arg>--workingPath</arg><arg>${workingPath}/publication</arg>
<arg>--pathMap</arg><arg>${pathMap}</arg> <arg>--pathMap</arg><arg>${pathMap}</arg>
<arg>--isLookUpUrl</arg><arg>${isLookUpUrl}</arg> <arg>--isLookUpUrl</arg><arg>${isLookUpUrl}</arg>
</spark> </spark>
@ -142,7 +89,7 @@
</spark-opts> </spark-opts>
<arg>--sourcePath</arg><arg>${sourcePath}/dataset</arg> <arg>--sourcePath</arg><arg>${sourcePath}/dataset</arg>
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg> <arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
<arg>--outputPath</arg><arg>${outputPath}/dataset</arg> <arg>--workingPath</arg><arg>${workingPath}/dataset</arg>
<arg>--pathMap</arg><arg>${pathMap}</arg> <arg>--pathMap</arg><arg>${pathMap}</arg>
<arg>--isLookUpUrl</arg><arg>${isLookUpUrl}</arg> <arg>--isLookUpUrl</arg><arg>${isLookUpUrl}</arg>
</spark> </spark>
@ -169,7 +116,7 @@
</spark-opts> </spark-opts>
<arg>--sourcePath</arg><arg>${sourcePath}/otherresearchproduct</arg> <arg>--sourcePath</arg><arg>${sourcePath}/otherresearchproduct</arg>
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg> <arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
<arg>--outputPath</arg><arg>${outputPath}/otherresearchproduct</arg> <arg>--workingPath</arg><arg>${workingPath}/otherresearchproduct</arg>
<arg>--pathMap</arg><arg>${pathMap}</arg> <arg>--pathMap</arg><arg>${pathMap}</arg>
<arg>--isLookUpUrl</arg><arg>${isLookUpUrl}</arg> <arg>--isLookUpUrl</arg><arg>${isLookUpUrl}</arg>
</spark> </spark>
@ -196,7 +143,7 @@
</spark-opts> </spark-opts>
<arg>--sourcePath</arg><arg>${sourcePath}/software</arg> <arg>--sourcePath</arg><arg>${sourcePath}/software</arg>
<arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg> <arg>--resultTableName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
<arg>--outputPath</arg><arg>${outputPath}/software</arg> <arg>--workingPath</arg><arg>${workingPath}/software</arg>
<arg>--pathMap</arg><arg>${pathMap}</arg> <arg>--pathMap</arg><arg>${pathMap}</arg>
<arg>--isLookUpUrl</arg><arg>${isLookUpUrl}</arg> <arg>--isLookUpUrl</arg><arg>${isLookUpUrl}</arg>
</spark> </spark>

View File

@ -36,49 +36,7 @@
<path start="copy_datasources"/> <path start="copy_datasources"/>
</fork> </fork>
<action name="copy_relation">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/relation</arg>
<arg>${nameNode}/${outputPath}/relation</arg>
</distcp>
<ok to="copy_wait"/>
<error to="Kill"/>
</action>
<action name="copy_organization">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/organization</arg>
<arg>${nameNode}/${outputPath}/organization</arg>
</distcp>
<ok to="copy_wait"/>
<error to="Kill"/>
</action>
<action name="copy_projects">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/project</arg>
<arg>${nameNode}/${outputPath}/project</arg>
</distcp>
<ok to="copy_wait"/>
<error to="Kill"/>
</action>
<action name="copy_datasources">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<job-tracker>${jobTracker}</job-tracker>
<name-node>${nameNode}</name-node>
<arg>${nameNode}/${sourcePath}/datasource</arg>
<arg>${nameNode}/${outputPath}/datasource</arg>
</distcp>
<ok to="copy_wait"/>
<error to="Kill"/>
</action>
<join name="copy_wait" to="fork_prepare_assoc_step1"/> <join name="copy_wait" to="fork_prepare_assoc_step1"/>

View File

@ -21,105 +21,12 @@
</configuration> </configuration>
</global> </global>
<start to="reset_outputpath"/> <start to="prepare_result_organization_association"/>
<kill name="Kill"> <kill name="Kill">
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message> <message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill> </kill>
<action name="reset_outputpath">
<fs>
<delete path="${outputPath}"/>
<mkdir path="${outputPath}"/>
</fs>
<ok to="copy_entities"/>
<error to="Kill"/>
</action>
<fork name="copy_entities">
<path start="copy_relation"/>
<path start="copy_publication"/>
<path start="copy_dataset"/>
<path start="copy_orp"/>
<path start="copy_software"/>
<path start="copy_organization"/>
<path start="copy_projects"/>
<path start="copy_datasources"/>
</fork>
<action name="copy_relation">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>${nameNode}/${sourcePath}/relation</arg>
<arg>${nameNode}/${outputPath}/relation</arg>
</distcp>
<ok to="wait"/>
<error to="Kill"/>
</action>
<action name="copy_publication">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>${nameNode}/${sourcePath}/publication</arg>
<arg>${nameNode}/${outputPath}/publication</arg>
</distcp>
<ok to="wait"/>
<error to="Kill"/>
</action>
<action name="copy_dataset">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>${nameNode}/${sourcePath}/dataset</arg>
<arg>${nameNode}/${outputPath}/dataset</arg>
</distcp>
<ok to="wait"/>
<error to="Kill"/>
</action>
<action name="copy_orp">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>${nameNode}/${sourcePath}/otherresearchproduct</arg>
<arg>${nameNode}/${outputPath}/otherresearchproduct</arg>
</distcp>
<ok to="wait"/>
<error to="Kill"/>
</action>
<action name="copy_software">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>${nameNode}/${sourcePath}/software</arg>
<arg>${nameNode}/${outputPath}/software</arg>
</distcp>
<ok to="wait"/>
<error to="Kill"/>
</action>
<action name="copy_organization">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>${nameNode}/${sourcePath}/organization</arg>
<arg>${nameNode}/${outputPath}/organization</arg>
</distcp>
<ok to="wait"/>
<error to="Kill"/>
</action>
<action name="copy_projects">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>${nameNode}/${sourcePath}/project</arg>
<arg>${nameNode}/${outputPath}/project</arg>
</distcp>
<ok to="wait"/>
<error to="Kill"/>
</action>
<action name="copy_datasources">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>${nameNode}/${sourcePath}/datasource</arg>
<arg>${nameNode}/${outputPath}/datasource</arg>
</distcp>
<ok to="wait"/>
<error to="Kill"/>
</action>
<join name="wait" to="prepare_result_organization_association"/>
<action name="prepare_result_organization_association"> <action name="prepare_result_organization_association">
<spark xmlns="uri:oozie:spark-action:0.2"> <spark xmlns="uri:oozie:spark-action:0.2">