2019-04-11 15:39:29 +02:00
|
|
|
<workflow-app name="CollectionWorkflow" xmlns="uri:oozie:workflow:0.5">
|
2019-03-25 15:18:31 +01:00
|
|
|
<parameters>
|
|
|
|
<property>
|
|
|
|
<name>apiDescription</name>
|
|
|
|
<description>A json encoding of the API Description class</description>
|
|
|
|
</property>
|
|
|
|
|
|
|
|
<property>
|
|
|
|
<name>dataSourceInfo</name>
|
|
|
|
<description>A json encoding of the Datasource Info</description>
|
|
|
|
</property>
|
|
|
|
<property>
|
2019-04-11 15:39:29 +02:00
|
|
|
<name>identifierPath</name>
|
2021-01-29 16:42:41 +01:00
|
|
|
<description>An xpath to retrieve the metadata identifier for the generation of DNet Identifier </description>
|
2019-03-25 15:18:31 +01:00
|
|
|
</property>
|
|
|
|
|
|
|
|
<property>
|
|
|
|
<name>metadataEncoding</name>
|
|
|
|
<description> The type of the metadata XML/JSON</description>
|
|
|
|
</property>
|
|
|
|
|
|
|
|
<property>
|
|
|
|
<name>timestamp</name>
|
|
|
|
<description>The timestamp of the collection date</description>
|
|
|
|
</property>
|
2019-04-11 15:39:29 +02:00
|
|
|
|
|
|
|
<property>
|
|
|
|
<name>workflowId</name>
|
|
|
|
<description>The identifier of the workflow</description>
|
|
|
|
</property>
|
2021-01-29 16:42:41 +01:00
|
|
|
|
|
|
|
<property>
|
|
|
|
<name>mdStoreID</name>
|
|
|
|
<description>The identifier of the mdStore</description>
|
|
|
|
</property>
|
|
|
|
|
|
|
|
<property>
|
|
|
|
<name>mdStoreManagerURI</name>
|
|
|
|
<description>The URI of the MDStore Manager</description>
|
|
|
|
</property>
|
|
|
|
|
|
|
|
<property>
|
|
|
|
<name>collectionMode</name>
|
2021-01-29 17:21:42 +01:00
|
|
|
<description>Should be REFRESH or INCREMENTAL</description>
|
2021-01-29 16:42:41 +01:00
|
|
|
</property>
|
|
|
|
|
2019-03-25 15:18:31 +01:00
|
|
|
</parameters>
|
2021-01-28 09:51:17 +01:00
|
|
|
<global>
|
|
|
|
<job-tracker>${jobTracker}</job-tracker>
|
|
|
|
<name-node>${nameNode}</name-node>
|
|
|
|
</global>
|
2019-03-25 15:18:31 +01:00
|
|
|
|
2021-01-29 16:42:41 +01:00
|
|
|
|
2021-02-01 13:56:05 +01:00
|
|
|
<start to="collection_mode"/>
|
2019-03-18 10:44:35 +01:00
|
|
|
<kill name="Kill">
|
|
|
|
<message>Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
|
|
</kill>
|
2021-01-28 09:51:17 +01:00
|
|
|
|
2021-01-29 16:42:41 +01:00
|
|
|
|
|
|
|
<decision name="collection_mode">
|
|
|
|
<switch>
|
|
|
|
<case to="StartTransaction">${wf:conf('collectionMode') eq 'REFRESH'}</case>
|
|
|
|
<case to="BeginRead">${wf:conf('collectionMode') eq 'INCREMENTAL'}</case>
|
2021-02-01 13:56:05 +01:00
|
|
|
<default to="StartTransaction"/>
|
2021-01-29 16:42:41 +01:00
|
|
|
</switch>
|
|
|
|
</decision>
|
|
|
|
|
|
|
|
<action name="BeginRead">
|
|
|
|
<java>
|
|
|
|
<main-class>eu.dnetlib.dhp.aggregation.mdstore.MDStoreActionNode</main-class>
|
|
|
|
<arg>--action</arg><arg>READ_LOCK</arg>
|
|
|
|
<arg>--mdStoreID</arg><arg>${mdStoreID}</arg>
|
|
|
|
<arg>--mdStoreManagerURI</arg><arg>${mdStoreManagerURI}</arg>
|
|
|
|
<capture-output/>
|
|
|
|
</java>
|
|
|
|
<ok to="StartTransaction"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
<action name="StartTransaction">
|
|
|
|
<java>
|
|
|
|
<main-class>eu.dnetlib.dhp.aggregation.mdstore.MDStoreActionNode</main-class>
|
|
|
|
<arg>--action</arg><arg>NEW_VERSION</arg>
|
|
|
|
<arg>--mdStoreID</arg><arg>${mdStoreID}</arg>
|
|
|
|
<arg>--mdStoreManagerURI</arg><arg>${mdStoreManagerURI}</arg>
|
|
|
|
<capture-output/>
|
|
|
|
</java>
|
|
|
|
<ok to="CollectionWorker"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
2019-03-18 10:44:35 +01:00
|
|
|
<action name="CollectionWorker">
|
2019-10-24 11:36:59 +02:00
|
|
|
<java>
|
2021-01-28 09:51:17 +01:00
|
|
|
<main-class>eu.dnetlib.dhp.collection.worker.CollectorWorkerApplication</main-class>
|
|
|
|
<arg>--apidescriptor</arg><arg>${apiDescription}</arg>
|
|
|
|
<arg>--namenode</arg><arg>${nameNode}</arg>
|
2021-01-29 16:42:41 +01:00
|
|
|
<arg>--mdStoreVersion</arg><arg>${wf:actionData('StartTransaction')['mdStoreVersion']}</arg>
|
2019-10-24 11:36:59 +02:00
|
|
|
</java>
|
2019-04-11 15:39:29 +02:00
|
|
|
<ok to="GenerateNativeStoreSparkJob"/>
|
2021-02-01 13:56:05 +01:00
|
|
|
<error to="FailCollection"/>
|
2019-03-18 10:44:35 +01:00
|
|
|
</action>
|
2021-01-28 09:51:17 +01:00
|
|
|
|
2019-04-11 15:39:29 +02:00
|
|
|
<action name="GenerateNativeStoreSparkJob">
|
2019-03-18 10:44:35 +01:00
|
|
|
<spark xmlns="uri:oozie:spark-action:0.2">
|
|
|
|
<master>yarn</master>
|
|
|
|
<mode>cluster</mode>
|
2021-01-28 09:51:17 +01:00
|
|
|
<name>Generate Native MetadataStore</name>
|
2019-03-18 10:44:35 +01:00
|
|
|
<class>eu.dnetlib.dhp.collection.GenerateNativeStoreSparkJob</class>
|
2021-01-28 09:51:17 +01:00
|
|
|
<jar>dhp-aggregation-${projectVersion}.jar</jar>
|
|
|
|
<spark-opts>
|
|
|
|
--executor-memory=${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}${spark2EventLogDir}
|
|
|
|
</spark-opts>
|
|
|
|
<arg>--encoding</arg><arg>${metadataEncoding}</arg>
|
|
|
|
<arg>--dateOfCollection</arg><arg>${timestamp}</arg>
|
|
|
|
<arg>--provenance</arg><arg>${dataSourceInfo}</arg>
|
2019-03-25 15:18:31 +01:00
|
|
|
<arg>--xpath</arg><arg>${identifierPath}</arg>
|
2021-01-29 16:42:41 +01:00
|
|
|
<arg>--mdStoreVersion</arg><arg>${wf:actionData('StartTransaction')['mdStoreVersion']}</arg>
|
2021-02-01 13:56:05 +01:00
|
|
|
<arg>--readMdStoreVersion</arg><arg>${wf:actionData('BeginRead')['mdStoreReadLockVersion']}</arg>
|
2019-03-18 10:44:35 +01:00
|
|
|
</spark>
|
2021-01-29 16:42:41 +01:00
|
|
|
<ok to="collection_mode_end"/>
|
2021-02-01 13:56:05 +01:00
|
|
|
<error to="FailCollection"/>
|
2021-01-29 16:42:41 +01:00
|
|
|
</action>
|
|
|
|
|
|
|
|
|
|
|
|
<decision name="collection_mode_end">
|
|
|
|
<switch>
|
|
|
|
<case to="CommitVersion">${wf:conf('collectionMode') eq 'REFRESH'}</case>
|
|
|
|
<case to="EndRead">${wf:conf('collectionMode') eq 'INCREMENTAL'}</case>
|
2021-02-01 13:56:05 +01:00
|
|
|
<default to="CommitVersion"/>
|
2021-01-29 16:42:41 +01:00
|
|
|
</switch>
|
|
|
|
</decision>
|
|
|
|
|
|
|
|
<action name="EndRead">
|
|
|
|
<java>
|
|
|
|
<main-class>eu.dnetlib.dhp.aggregation.mdstore.MDStoreActionNode</main-class>
|
|
|
|
<arg>--action</arg><arg>READ_UNLOCK</arg>
|
|
|
|
<arg>--mdStoreManagerURI</arg><arg>${mdStoreManagerURI}</arg>
|
|
|
|
<arg>--readMDStoreId</arg><arg>${wf:actionData('BeginRead')['mdStoreReadLockVersion']}</arg>
|
|
|
|
<capture-output/>
|
|
|
|
</java>
|
|
|
|
<ok to="CommitVersion"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
<action name="CommitVersion">
|
|
|
|
<java>
|
|
|
|
<main-class>eu.dnetlib.dhp.aggregation.mdstore.MDStoreActionNode</main-class>
|
|
|
|
<arg>--action</arg><arg>COMMIT</arg>
|
|
|
|
<arg>--namenode</arg><arg>${nameNode}</arg>
|
|
|
|
<arg>--mdStoreVersion</arg><arg>${wf:actionData('StartTransaction')['mdStoreVersion']}</arg>
|
|
|
|
<arg>--mdStoreManagerURI</arg><arg>${mdStoreManagerURI}</arg>
|
|
|
|
</java>
|
2019-03-18 10:44:35 +01:00
|
|
|
<ok to="End"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
2019-04-11 15:39:29 +02:00
|
|
|
|
2021-02-01 13:56:05 +01:00
|
|
|
<decision name="FailCollection">
|
|
|
|
<switch>
|
|
|
|
<case to="RollBack">${wf:conf('collectionMode') eq 'REFRESH'}</case>
|
|
|
|
<case to="EndReadRollBack">${wf:conf('collectionMode') eq 'INCREMENTAL'}</case>
|
|
|
|
<default to="RollBack"/>
|
|
|
|
</switch>
|
|
|
|
</decision>
|
|
|
|
|
|
|
|
<action name="EndReadRollBack">
|
|
|
|
<java>
|
|
|
|
<main-class>eu.dnetlib.dhp.aggregation.mdstore.MDStoreActionNode</main-class>
|
|
|
|
<arg>--action</arg><arg>READ_UNLOCK</arg>
|
|
|
|
<arg>--mdStoreManagerURI</arg><arg>${mdStoreManagerURI}</arg>
|
|
|
|
<arg>--readMDStoreId</arg><arg>${wf:actionData('BeginRead')['mdStoreReadLockVersion']}</arg>
|
|
|
|
<capture-output/>
|
|
|
|
</java>
|
|
|
|
<ok to="RollBack"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-01-29 16:42:41 +01:00
|
|
|
<action name="RollBack">
|
|
|
|
<java>
|
|
|
|
<main-class>eu.dnetlib.dhp.aggregation.mdstore.MDStoreActionNode</main-class>
|
2021-01-29 17:21:42 +01:00
|
|
|
<arg>--action</arg><arg>ROLLBACK</arg>
|
|
|
|
<arg>--mdStoreVersion</arg><arg>${wf:actionData('StartTransaction')['mdStoreVersion']}</arg>
|
2021-01-29 16:42:41 +01:00
|
|
|
<arg>--mdStoreManagerURI</arg><arg>${mdStoreManagerURI}</arg>
|
|
|
|
</java>
|
|
|
|
<ok to="Kill"/>
|
|
|
|
<error to="Kill"/>
|
|
|
|
</action>
|
|
|
|
|
|
|
|
|
2019-03-18 10:44:35 +01:00
|
|
|
<end name="End"/>
|
|
|
|
</workflow-app>
|