101 lines
5.0 KiB
XML
101 lines
5.0 KiB
XML
<RESOURCE_PROFILE>
|
|
<HEADER>
|
|
<RESOURCE_IDENTIFIER value="7c8765af-1253-4bd7-8806-315b73bf7319_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
|
|
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
|
|
<RESOURCE_KIND value="WorkflowDSResources"/>
|
|
<RESOURCE_URI value=""/>
|
|
<DATE_OF_CREATION value="2021-07-15T16:06:50+00:00"/>
|
|
</HEADER>
|
|
<BODY>
|
|
<WORKFLOW_NAME>Import Orcid</WORKFLOW_NAME>
|
|
<WORKFLOW_TYPE>Import InfoSpace</WORKFLOW_TYPE>
|
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
|
<CONFIGURATION start="manual">
|
|
<NODE isStart="true" name="setInputPath" type="SetEnvParameter">
|
|
<DESCRIPTION>set the hdfs input path</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">inputPath</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/orcid_activities_2020</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="setProcessOutputPath"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="setProcessOutputPath" type="SetEnvParameter">
|
|
<DESCRIPTION>set the temporary path where to store the action set</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">processOutputPath</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/prod_provision/working_path_orcid_activities</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="prepareActionSets"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="prepareActionSets" type="PrepareActionSets">
|
|
<DESCRIPTION>prepare action sets</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="sets" required="true" type="string">
|
|
[
|
|
{
|
|
'set' : 'orcidworks-no-doi',
|
|
'jobProperty' : 'export_action_set_orcidworks_no_doi',
|
|
'enablingProperty' : 'active_orcidworks_no_doi',
|
|
'enabled' : 'true'
|
|
}
|
|
]
|
|
</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="extractOutputPath"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="extractOutputPath" type="ExtractOutputPath">
|
|
<DESCRIPTION>extract the hdfs output path generated in the previous node</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="hdfsOutputPathParam" required="true" type="string">outputPath</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="orcidNoDoiUpdate"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="orcidNoDoiUpdate" type="SubmitHadoopJob">
|
|
<DESCRIPTION>prepare updates for the Orcid No Doi</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
|
|
<PARAM managedBy="system" name="cluster" required="true" type="string">IIS</PARAM>
|
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
|
{
|
|
'workingPath' : 'inputPath',
|
|
'processOutputPath' : 'processOutputPath',
|
|
'outputPath': 'outputPath'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="params" required="true" type="string">
|
|
{
|
|
'oozie.wf.application.path' : '/lib/dnet/PROD/actionmanager/orcidnodoi_actionset/oozie_app',
|
|
'spark2GenNoDoiDatasetMaxExecutors' : '200',
|
|
'spark2GenNoDoiDatasetExecutorMemory' : '2G'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="updateActionSets"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="updateActionSets" type="UpdateActionSets">
|
|
<DESCRIPTION>update action sets</DESCRIPTION>
|
|
<PARAMETERS/>
|
|
<ARCS>
|
|
<ARC to="success"/>
|
|
</ARCS>
|
|
</NODE>
|
|
</CONFIGURATION>
|
|
<STATUS>
|
|
<LAST_EXECUTION_ID>wf_20210713_170819_470</LAST_EXECUTION_ID>
|
|
<LAST_EXECUTION_DATE>2021-07-13T17:28:26+00:00</LAST_EXECUTION_DATE>
|
|
<LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
|
|
<LAST_EXECUTION_ERROR/>
|
|
</STATUS>
|
|
</BODY>
|
|
</RESOURCE_PROFILE> |