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

[dhp-actionmanager] promoting workflow added

This commit is contained in:
Przemysław Jacewicz 2020-04-01 18:51:25 +02:00 committed by przemek
parent 5b459bcc47
commit 80cf43b9c8
10 changed files with 1485 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

178
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/dataset/oozie_app/workflow.xml Normal file
View File

@ -0,0 +1,178 @@
<workflow-app name="promote_action_payload_for_dataset_table" xmlns="uri:oozie:workflow:0.5">
<parameters>
<property>
<name>activePromoteDatasetActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>activePromoteResultActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>inputGraphPath</name>
<description>TODO</description>
</property>
<property>
<name>inputActionPayloadRootPath</name>
<description>TODO</description>
</property>
<property>
<name>outputGraphPath</name>
<description>TODO</description>
</property>
<property>
<name>mergeAndGetStrategy</name>
<description>TODO</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>oozie.action.sharelib.for.spark</name>
<value>${oozieActionShareLibForSpark2}</value>
</property>
</configuration>
</global>
<start to="DecisionPromoteDatasetActionPayload"/>
<kill name="Kill">
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill>
<decision name="DecisionPromoteDatasetActionPayload">
<switch>
<case to="PromoteDatasetActionPayloadForDatasetTable">
${(activePromoteDatasetActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphPath')),'/'),'dataset')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.Dataset')) eq "true")}
</case>
<default to="SkipPromoteDatasetActionPayloadForDatasetTable"/>
</switch>
</decision>
<action name="PromoteDatasetActionPayloadForDatasetTable">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>PromoteDatasetActionPayloadForDatasetTable</name>
<class>eu.dnetlib.dhp.actionmanager.promote.PromoteActionPayloadForGraphTableJob</class>
<jar>dhp-actionmanager-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=2560
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${inputGraphPath}/dataset</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Dataset</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
<arg>--outputGraphTablePath</arg><arg>${workingDir}/dataset</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="DecisionPromoteResultActionPayloadForDatasetTable"/>
<error to="Kill"/>
</action>
<action name="SkipPromoteDatasetActionPayloadForDatasetTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${inputGraphPath}/dataset</arg>
<arg>${workingDir}/dataset</arg>
</distcp>
<ok to="DecisionPromoteResultActionPayloadForDatasetTable"/>
<error to="Kill"/>
</action>
<decision name="DecisionPromoteResultActionPayloadForDatasetTable">
<switch>
<case to="PromoteResultActionPayloadForDatasetTable">
${(activePromoteResultActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.Result')) eq "true")}
</case>
<default to="SkipPromoteResultActionPayloadForDatasetTable"/>
</switch>
</decision>
<action name="PromoteResultActionPayloadForDatasetTable">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>PromoteResultActionPayloadForDatasetTable</name>
<class>eu.dnetlib.dhp.actionmanager.promote.PromoteActionPayloadForGraphTableJob</class>
<jar>dhp-actionmanager-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=2560
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${workingDir}/dataset</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Dataset</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphPath}/dataset</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="End"/>
<error to="Kill"/>
</action>
<action name="SkipPromoteResultActionPayloadForDatasetTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${workingDir}/dataset</arg>
<arg>${outputGraphPath}/dataset</arg>
</distcp>
<ok to="End"/>
<error to="Kill"/>
</action>
<end name="End"/>
</workflow-app>

125
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/datasource/oozie_app/workflow.xml Normal file
View File

@ -0,0 +1,125 @@
<workflow-app name="promote_action_payload_for_datasource_table" xmlns="uri:oozie:workflow:0.5">
<parameters>
<property>
<name>activePromoteDatasourceActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>inputGraphPath</name>
<description>TODO</description>
</property>
<property>
<name>inputActionPayloadRootPath</name>
<description>TODO</description>
</property>
<property>
<name>outputGraphPath</name>
<description>TODO</description>
</property>
<property>
<name>mergeAndGetStrategy</name>
<description>TODO</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>oozie.action.sharelib.for.spark</name>
<value>${oozieActionShareLibForSpark2}</value>
</property>
</configuration>
</global>
<start to="DecisionPromoteDatasourceActionPayload"/>
<kill name="Kill">
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill>
<decision name="DecisionPromoteDatasourceActionPayload">
<switch>
<case to="PromoteDatasourceActionPayloadForDatasourceTable">
${(activePromoteDatasourceActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphPath')),'/'),'datasource')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.Datasource')) eq "true")}
</case>
<default to="SkipPromoteDatasourceActionPayloadForDatasourceTable"/>
</switch>
</decision>
<action name="PromoteDatasourceActionPayloadForDatasourceTable">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>PromoteDatasourceActionPayloadForDatasourceTable</name>
<class>eu.dnetlib.dhp.actionmanager.promote.PromoteActionPayloadForGraphTableJob</class>
<jar>dhp-actionmanager-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${inputGraphPath}/datasource</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Datasource</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Datasource</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Datasource</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphPath}/datasource</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="End"/>
<error to="Kill"/>
</action>
<action name="SkipPromoteDatasourceActionPayloadForDatasourceTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${inputGraphPath}/datasource</arg>
<arg>${outputGraphPath}/datasource</arg>
</distcp>
<ok to="End"/>
<error to="Kill"/>
</action>
<end name="End"/>
</workflow-app>

9
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/main/oozie_app/import.txt Normal file
View File

@ -0,0 +1,9 @@
## This is a classpath-based import file (this header is required)
promote_action_payload_for_dataset_table classpath eu/dnetlib/dhp/actionmanager/wf/dataset/oozie_app
promote_action_payload_for_datasource_table classpath eu/dnetlib/dhp/actionmanager/wf/datasource/oozie_app
promote_action_payload_for_organization_table classpath eu/dnetlib/dhp/actionmanager/wf/organization/oozie_app
promote_action_payload_for_otherresearchproduct_table classpath eu/dnetlib/dhp/actionmanager/wf/otherresearchproduct/oozie_app
promote_action_payload_for_project_table classpath eu/dnetlib/dhp/actionmanager/wf/project/oozie_app
promote_action_payload_for_publication_table classpath eu/dnetlib/dhp/actionmanager/wf/publication/oozie_app
promote_action_payload_for_relation_table classpath eu/dnetlib/dhp/actionmanager/wf/relation/oozie_app
promote_action_payload_for_software_table classpath eu/dnetlib/dhp/actionmanager/wf/software/oozie_app

265
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/main/oozie_app/workflow.xml Normal file
View File

@ -0,0 +1,265 @@
<workflow-app name="promote_actionsets_for_graph" xmlns="uri:oozie:workflow:0.5">
<parameters>
<property>
<name>activePromoteDatasetActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>activePromoteDatasourceActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>activePromoteOrganizationActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>activePromoteOtherResearchProductActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>activePromoteProjectActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>activePromotePublicationActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>activePromoteRelationActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>activePromoteResultActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>activePromoteSoftwareActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>inputGraphPath</name>
<description>TODO</description>
</property>
<property>
<name>inputActionSetPaths</name>
<description>TODO</description>
</property>
<property>
<name>outputGraphPath</name>
<description>TODO</description>
</property>
<property>
<name>mergeAndGetStrategy</name>
<description>TODO</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>oozie.action.sharelib.for.spark</name>
<value>${oozieActionShareLibForSpark2}</value>
</property>
</configuration>
</global>
<start to="PartitionActionSetsByPayloadType"/>
<kill name="Kill">
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill>
<action name="PartitionActionSetsByPayloadType">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>PartitionActionSetsByPayloadType</name>
<class>eu.dnetlib.dhp.actionmanager.partition.PartitionActionSetsByPayloadTypeJob</class>
<jar>dhp-actionmanager-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--inputActionSetPaths</arg><arg>${inputActionSetPaths}</arg>
<arg>--outputPath</arg><arg>${workingDir}/action_payload_by_type</arg>
</spark>
<ok to="ForkPromote"/>
<error to="Kill"/>
</action>
<fork name="ForkPromote">
<path start="PromoteActionPayloadForDatasetTable"/>
<path start="PromoteActionPayloadForDatasourceTable"/>
<path start="PromoteActionPayloadForOrganizationTable"/>
<path start="PromoteActionPayloadForOtherResearchProductTable"/>
<path start="PromoteActionPayloadForProjectTable"/>
<path start="PromoteActionPayloadForPublicationTable"/>
<path start="PromoteActionPayloadForRelationTable"/>
<path start="PromoteActionPayloadForSoftwareTable"/>
</fork>
<action name="PromoteActionPayloadForDatasetTable">
<sub-workflow>
<app-path>${wf:appPath()}/promote_action_payload_for_dataset_table</app-path>
<propagate-configuration/>
<configuration>
<property>
<name>inputActionPayloadRootPath</name>
<value>${workingDir}/action_payload_by_type</value>
</property>
</configuration>
</sub-workflow>
<ok to="JoinPromote"/>
<error to="Kill"/>
</action>
<action name="PromoteActionPayloadForDatasourceTable">
<sub-workflow>
<app-path>${wf:appPath()}/promote_action_payload_for_datasource_table</app-path>
<propagate-configuration/>
<configuration>
<property>
<name>inputActionPayloadRootPath</name>
<value>${workingDir}/action_payload_by_type</value>
</property>
</configuration>
</sub-workflow>
<ok to="JoinPromote"/>
<error to="Kill"/>
</action>
<action name="PromoteActionPayloadForOrganizationTable">
<sub-workflow>
<app-path>${wf:appPath()}/promote_action_payload_for_organization_table</app-path>
<propagate-configuration/>
<configuration>
<property>
<name>inputActionPayloadRootPath</name>
<value>${workingDir}/action_payload_by_type</value>
</property>
</configuration>
</sub-workflow>
<ok to="JoinPromote"/>
<error to="Kill"/>
</action>
<action name="PromoteActionPayloadForOtherResearchProductTable">
<sub-workflow>
<app-path>${wf:appPath()}/promote_action_payload_for_otherresearchproduct_table</app-path>
<propagate-configuration/>
<configuration>
<property>
<name>inputActionPayloadRootPath</name>
<value>${workingDir}/action_payload_by_type</value>
</property>
</configuration>
</sub-workflow>
<ok to="JoinPromote"/>
<error to="Kill"/>
</action>
<action name="PromoteActionPayloadForProjectTable">
<sub-workflow>
<app-path>${wf:appPath()}/promote_action_payload_for_project_table</app-path>
<propagate-configuration/>
<configuration>
<property>
<name>inputActionPayloadRootPath</name>
<value>${workingDir}/action_payload_by_type</value>
</property>
</configuration>
</sub-workflow>
<ok to="JoinPromote"/>
<error to="Kill"/>
</action>
<action name="PromoteActionPayloadForPublicationTable">
<sub-workflow>
<app-path>${wf:appPath()}/promote_action_payload_for_publication_table</app-path>
<propagate-configuration/>
<configuration>
<property>
<name>inputActionPayloadRootPath</name>
<value>${workingDir}/action_payload_by_type</value>
</property>
</configuration>
</sub-workflow>
<ok to="JoinPromote"/>
<error to="Kill"/>
</action>
<action name="PromoteActionPayloadForRelationTable">
<sub-workflow>
<app-path>${wf:appPath()}/promote_action_payload_for_relation_table</app-path>
<propagate-configuration/>
<configuration>
<property>
<name>inputActionPayloadRootPath</name>
<value>${workingDir}/action_payload_by_type</value>
</property>
</configuration>
</sub-workflow>
<ok to="JoinPromote"/>
<error to="Kill"/>
</action>
<action name="PromoteActionPayloadForSoftwareTable">
<sub-workflow>
<app-path>${wf:appPath()}/promote_action_payload_for_software_table</app-path>
<propagate-configuration/>
<configuration>
<property>
<name>inputActionPayloadRootPath</name>
<value>${workingDir}/action_payload_by_type</value>
</property>
</configuration>
</sub-workflow>
<ok to="JoinPromote"/>
<error to="Kill"/>
</action>
<join name="JoinPromote" to="End"/>
<end name="End"/>
</workflow-app>

125
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/organization/oozie_app/workflow.xml Normal file
View File

@ -0,0 +1,125 @@
<workflow-app name="promote_action_payload_for_organization_table" xmlns="uri:oozie:workflow:0.5">
<parameters>
<property>
<name>activePromoteOrganizationActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>inputGraphPath</name>
<description>TODO</description>
</property>
<property>
<name>inputActionPayloadRootPath</name>
<description>TODO</description>
</property>
<property>
<name>outputGraphPath</name>
<description>TODO</description>
</property>
<property>
<name>mergeAndGetStrategy</name>
<description>TODO</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>oozie.action.sharelib.for.spark</name>
<value>${oozieActionShareLibForSpark2}</value>
</property>
</configuration>
</global>
<start to="DecisionPromoteOrganizationActionPayload"/>
<kill name="Kill">
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill>
<decision name="DecisionPromoteOrganizationActionPayload">
<switch>
<case to="PromoteOrganizationActionPayloadForOrganizationTable">
${(activePromoteOrganizationActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphPath')),'/'),'organization')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.Organization')) eq "true")}
</case>
<default to="SkipPromoteOrganizationActionPayloadForOrganizationTable"/>
</switch>
</decision>
<action name="PromoteOrganizationActionPayloadForOrganizationTable">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>PromoteOrganizationActionPayloadForOrganizationTable</name>
<class>eu.dnetlib.dhp.actionmanager.promote.PromoteActionPayloadForGraphTableJob</class>
<jar>dhp-actionmanager-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${inputGraphPath}/organization</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Organization</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Organization</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Organization</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphPath}/organization</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="End"/>
<error to="Kill"/>
</action>
<action name="SkipPromoteOrganizationActionPayloadForOrganizationTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${inputGraphPath}/organization</arg>
<arg>${outputGraphPath}/organization</arg>
</distcp>
<ok to="End"/>
<error to="Kill"/>
</action>
<end name="End"/>
</workflow-app>

177
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/otherresearchproduct/oozie_app/workflow.xml Normal file
View File

@ -0,0 +1,177 @@
<workflow-app name="promote_action_payload_for_otherresearchproduct_table" xmlns="uri:oozie:workflow:0.5">
<parameters>
<property>
<name>activePromoteOtherResearchProductActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>activePromoteResultActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>inputGraphPath</name>
<description>TODO</description>
</property>
<property>
<name>inputActionPayloadRootPath</name>
<description>TODO</description>
</property>
<property>
<name>outputGraphPath</name>
<description>TODO</description>
</property>
<property>
<name>mergeAndGetStrategy</name>
<description>TODO</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>oozie.action.sharelib.for.spark</name>
<value>${oozieActionShareLibForSpark2}</value>
</property>
</configuration>
</global>
<start to="DecisionPromoteOtherResearchProductActionPayload"/>
<kill name="Kill">
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill>
<decision name="DecisionPromoteOtherResearchProductActionPayload">
<switch>
<case to="PromoteOtherResearchProductActionPayloadForOtherResearchProductTable">
${(activePromoteOtherResearchProductActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphPath')),'/'),'otherresearchproduct')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.OtherResearchProduct')) eq "true")}
</case>
<default to="SkipPromoteOtherResearchProductActionPayloadForOtherResearchProductTable"/>
</switch>
</decision>
<action name="PromoteOtherResearchProductActionPayloadForOtherResearchProductTable">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>PromoteOtherResearchProductActionPayloadForOtherResearchProductTable</name>
<class>eu.dnetlib.dhp.actionmanager.promote.PromoteActionPayloadForGraphTableJob</class>
<jar>dhp-actionmanager-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${inputGraphPath}/otherresearchproduct</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
<arg>--outputGraphTablePath</arg><arg>${workingDir}/otherresearchproduct</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="DecisionPromoteResultActionPayloadForOtherResearchProductTable"/>
<error to="Kill"/>
</action>
<action name="SkipPromoteOtherResearchProductActionPayloadForOtherResearchProductTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${inputGraphPath}/otherresearchproduct</arg>
<arg>${workingDir}/otherresearchproduct</arg>
</distcp>
<ok to="DecisionPromoteResultActionPayloadForOtherResearchProductTable"/>
<error to="Kill"/>
</action>
<decision name="DecisionPromoteResultActionPayloadForOtherResearchProductTable">
<switch>
<case to="PromoteResultActionPayloadForOtherResearchProductTable">
${(activePromoteResultActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.Result')) eq "true")}
</case>
<default to="SkipPromoteResultActionPayloadForOtherResearchProductTable"/>
</switch>
</decision>
<action name="PromoteResultActionPayloadForOtherResearchProductTable">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>PromoteResultActionPayloadForOtherResearchProductTable</name>
<class>eu.dnetlib.dhp.actionmanager.promote.PromoteActionPayloadForGraphTableJob</class>
<jar>dhp-actionmanager-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=2560
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${workingDir}/otherresearchproduct</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.OtherResearchProduct</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphPath}/otherresearchproduct</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="End"/>
<error to="Kill"/>
</action>
<action name="SkipPromoteResultActionPayloadForOtherResearchProductTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${workingDir}/otherresearchproduct</arg>
<arg>${outputGraphPath}/otherresearchproduct</arg>
</distcp>
<ok to="End"/>
<error to="Kill"/>
</action>
<end name="End"/>
</workflow-app>

125
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/project/oozie_app/workflow.xml Normal file
View File

@ -0,0 +1,125 @@
<workflow-app name="promote_action_payload_for_project_table" xmlns="uri:oozie:workflow:0.5">
<parameters>
<property>
<name>activePromoteProjectActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>inputGraphPath</name>
<description>TODO</description>
</property>
<property>
<name>inputActionPayloadRootPath</name>
<description>TODO</description>
</property>
<property>
<name>outputGraphPath</name>
<description>TODO</description>
</property>
<property>
<name>mergeAndGetStrategy</name>
<description>TODO</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>oozie.action.sharelib.for.spark</name>
<value>${oozieActionShareLibForSpark2}</value>
</property>
</configuration>
</global>
<start to="DecisionPromoteProjectActionPayload"/>
<kill name="Kill">
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill>
<decision name="DecisionPromoteProjectActionPayload">
<switch>
<case to="PromoteProjectActionPayloadForProjectTable">
${(activePromoteProjectActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphPath')),'/'),'project')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.Project')) eq "true")}
</case>
<default to="SkipPromoteProjectActionPayloadForProjectTable"/>
</switch>
</decision>
<action name="PromoteProjectActionPayloadForProjectTable">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>PromoteProjectActionPayloadForProjectTable</name>
<class>eu.dnetlib.dhp.actionmanager.promote.PromoteActionPayloadForGraphTableJob</class>
<jar>dhp-actionmanager-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${inputGraphPath}/project</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Project</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Project</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Project</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphPath}/project</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="End"/>
<error to="Kill"/>
</action>
<action name="SkipPromoteProjectActionPayloadForProjectTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${inputGraphPath}/project</arg>
<arg>${outputGraphPath}/project</arg>
</distcp>
<ok to="End"/>
<error to="Kill"/>
</action>
<end name="End"/>
</workflow-app>

178
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/publication/oozie_app/workflow.xml Normal file
View File

@ -0,0 +1,178 @@
<workflow-app name="promote_action_payload_for_publication_table" xmlns="uri:oozie:workflow:0.5">
<parameters>
<property>
<name>activePromotePublicationActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>activePromoteResultActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>inputGraphPath</name>
<description>TODO</description>
</property>
<property>
<name>inputActionPayloadRootPath</name>
<description>TODO</description>
</property>
<property>
<name>outputGraphPath</name>
<description>TODO</description>
</property>
<property>
<name>mergeAndGetStrategy</name>
<description>TODO</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>oozie.action.sharelib.for.spark</name>
<value>${oozieActionShareLibForSpark2}</value>
</property>
</configuration>
</global>
<start to="DecisionPromotePublicationActionPayload"/>
<kill name="Kill">
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill>
<decision name="DecisionPromotePublicationActionPayload">
<switch>
<case to="PromotePublicationActionPayloadForPublicationTable">
${(activePromotePublicationActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphPath')),'/'),'publication')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.Publication')) eq "true")}
</case>
<default to="SkipPromotePublicationActionPayloadForPublicationTable"/>
</switch>
</decision>
<action name="PromotePublicationActionPayloadForPublicationTable">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>PromotePublicationActionPayloadForPublicationTable</name>
<class>eu.dnetlib.dhp.actionmanager.promote.PromoteActionPayloadForGraphTableJob</class>
<jar>dhp-actionmanager-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=2560
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${inputGraphPath}/publication</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Publication</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
<arg>--outputGraphTablePath</arg><arg>${workingDir}/publication</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="DecisionPromoteResultActionPayloadForPublicationTable"/>
<error to="Kill"/>
</action>
<action name="SkipPromotePublicationActionPayloadForPublicationTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${inputGraphPath}/publication</arg>
<arg>${workingDir}/publication</arg>
</distcp>
<ok to="DecisionPromoteResultActionPayloadForPublicationTable"/>
<error to="Kill"/>
</action>
<decision name="DecisionPromoteResultActionPayloadForPublicationTable">
<switch>
<case to="PromoteResultActionPayloadForPublicationTable">
${(activePromoteResultActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.Result')) eq "true")}
</case>
<default to="SkipPromoteResultActionPayloadForPublicationTable"/>
</switch>
</decision>
<action name="PromoteResultActionPayloadForPublicationTable">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>PromoteResultActionPayloadForPublicationTable</name>
<class>eu.dnetlib.dhp.actionmanager.promote.PromoteActionPayloadForGraphTableJob</class>
<jar>dhp-actionmanager-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=2560
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${workingDir}/publication</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Publication</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphPath}/publication</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="End"/>
<error to="Kill"/>
</action>
<action name="SkipPromoteResultActionPayloadForPublicationTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${workingDir}/publication</arg>
<arg>${outputGraphPath}/publication</arg>
</distcp>
<ok to="End"/>
<error to="Kill"/>
</action>
<end name="End"/>
</workflow-app>

126
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/relation/oozie_app/workflow.xml Normal file
View File

@ -0,0 +1,126 @@
<workflow-app name="promote_action_payload_for_relation_table" xmlns="uri:oozie:workflow:0.5">
<parameters>
<property>
<name>activePromoteRelationActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>inputGraphPath</name>
<description>TODO</description>
</property>
<property>
<name>inputActionPayloadRootPath</name>
<description>TODO</description>
</property>
<property>
<name>outputGraphPath</name>
<description>TODO</description>
</property>
<property>
<name>mergeAndGetStrategy</name>
<description>TODO</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>oozie.action.sharelib.for.spark</name>
<value>${oozieActionShareLibForSpark2}</value>
</property>
</configuration>
</global>
<start to="DecisionPromoteRelationActionPayload"/>
<kill name="Kill">
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill>
<decision name="DecisionPromoteRelationActionPayload">
<switch>
<case to="PromoteRelationActionPayloadForRelationTable">
${(activePromoteRelationActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphPath')),'/'),'relation')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.Relation')) eq "true")}
</case>
<default to="SkipPromoteRelationActionPayloadForRelationTable"/>
</switch>
</decision>
<action name="PromoteRelationActionPayloadForRelationTable">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>PromoteRelationActionPayloadForRelationTable</name>
<class>eu.dnetlib.dhp.actionmanager.promote.PromoteActionPayloadForGraphTableJob</class>
<jar>dhp-actionmanager-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=2560
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${inputGraphPath}/relation</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Relation</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Relation</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Relation</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphPath}/relation</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="End"/>
<error to="Kill"/>
</action>
<action name="SkipPromoteRelationActionPayloadForRelationTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${inputGraphPath}/relation</arg>
<arg>${outputGraphPath}/relation</arg>
</distcp>
<ok to="End"/>
<error to="Kill"/>
</action>
<end name="End"/>
</workflow-app>

177
dhp-workflows/dhp-actionmanager/src/main/resources/eu/dnetlib/dhp/actionmanager/wf/software/oozie_app/workflow.xml Normal file
View File

@ -0,0 +1,177 @@
<workflow-app name="promote_action_payload_for_software_table" xmlns="uri:oozie:workflow:0.5">
<parameters>
<property>
<name>activePromoteSoftwareActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>activePromoteResultActionPayload</name>
<description>TODO</description>
</property>
<property>
<name>inputGraphPath</name>
<description>TODO</description>
</property>
<property>
<name>inputActionPayloadRootPath</name>
<description>TODO</description>
</property>
<property>
<name>outputGraphPath</name>
<description>TODO</description>
</property>
<property>
<name>mergeAndGetStrategy</name>
<description>TODO</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>oozie.action.sharelib.for.spark</name>
<value>${oozieActionShareLibForSpark2}</value>
</property>
</configuration>
</global>
<start to="DecisionPromoteSoftfwareActionPayload"/>
<kill name="Kill">
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
</kill>
<decision name="DecisionPromoteSoftfwareActionPayload">
<switch>
<case to="PromoteSoftwareActionPayloadForSoftwareTable">
${(activePromoteSoftwareActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputGraphPath')),'/'),'software')) eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.Software')) eq "true")}
</case>
<default to="SkipPromoteSoftwareActionPayloadForSoftwareTable"/>
</switch>
</decision>
<action name="PromoteSoftwareActionPayloadForSoftwareTable">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>PromoteSoftwareActionPayloadForSoftwareTable</name>
<class>eu.dnetlib.dhp.actionmanager.promote.PromoteActionPayloadForGraphTableJob</class>
<jar>dhp-actionmanager-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${inputGraphPath}/software</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Software</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
<arg>--outputGraphTablePath</arg><arg>${workingDir}/software</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="DecisionPromoteResultActionPayloadForSoftwareTable"/>
<error to="Kill"/>
</action>
<action name="SkipPromoteSoftwareActionPayloadForSoftwareTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${inputGraphPath}/software</arg>
<arg>${workingDir}/software</arg>
</distcp>
<ok to="DecisionPromoteResultActionPayloadForSoftwareTable"/>
<error to="Kill"/>
</action>
<decision name="DecisionPromoteResultActionPayloadForSoftwareTable">
<switch>
<case to="PromoteResultActionPayloadForSoftwareTable">
${(activePromoteResultActionPayload eq "true") and
(fs:exists(concat(concat(concat(concat(wf:conf('nameNode'),'/'),wf:conf('inputActionPayloadRootPath')),'/'),'clazz=eu.dnetlib.dhp.schema.oaf.Result')) eq "true")}
</case>
<default to="SkipPromoteResultActionPayloadForSoftwareTable"/>
</switch>
</decision>
<action name="PromoteResultActionPayloadForSoftwareTable">
<spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master>
<mode>cluster</mode>
<name>PromoteResultActionPayloadForSoftwareTable</name>
<class>eu.dnetlib.dhp.actionmanager.promote.PromoteActionPayloadForGraphTableJob</class>
<jar>dhp-actionmanager-${projectVersion}.jar</jar>
<spark-opts>
--executor-memory=${sparkExecutorMemory}
--executor-cores=${sparkExecutorCores}
--driver-memory=${sparkDriverMemory}
--conf spark.extraListeners=${spark2ExtraListeners}
--conf spark.sql.queryExecutionListeners=${spark2SqlQueryExecutionListeners}
--conf spark.yarn.historyServer.address=${spark2YarnHistoryServerAddress}
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.shuffle.partitions=2560
</spark-opts>
<arg>--inputGraphTablePath</arg><arg>${workingDir}/software</arg>
<arg>--graphTableClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Software</arg>
<arg>--inputActionPayloadPath</arg><arg>${inputActionPayloadRootPath}/clazz=eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--actionPayloadClassName</arg><arg>eu.dnetlib.dhp.schema.oaf.Result</arg>
<arg>--outputGraphTablePath</arg><arg>${outputGraphPath}/software</arg>
<arg>--mergeAndGetStrategy</arg><arg>${mergeAndGetStrategy}</arg>
</spark>
<ok to="End"/>
<error to="Kill"/>
</action>
<action name="SkipPromoteResultActionPayloadForSoftwareTable">
<distcp xmlns="uri:oozie:distcp-action:0.2">
<arg>-pb</arg>
<arg>${workingDir}/software</arg>
<arg>${outputGraphPath}/software</arg>
</distcp>
<ok to="End"/>
<error to="Kill"/>
</action>
<end name="End"/>
</workflow-app>