132 lines
6.7 KiB
XML
132 lines
6.7 KiB
XML
<RESOURCE_PROFILE>
|
|
<HEADER>
|
|
<RESOURCE_IDENTIFIER value="4bb067d5-a2f2-42b9-844c-4e1d8d71b80f_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
|
|
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
|
|
<RESOURCE_KIND value="WorkflowDSResources"/>
|
|
<RESOURCE_URI value=""/>
|
|
<DATE_OF_CREATION value="2021-05-20T15:00:27+00:00"/>
|
|
</HEADER>
|
|
<BODY>
|
|
<WORKFLOW_NAME>Import H2020classification</WORKFLOW_NAME>
|
|
<WORKFLOW_TYPE>Import H2020classification</WORKFLOW_TYPE>
|
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
|
<CONFIGURATION start="manual">
|
|
<NODE isStart="true" name="setProjectFileURL" type="SetEnvParameter">
|
|
<DESCRIPTION>sets the URL to download the project file</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">projectFileURL</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">https://cordis.europa.eu/data/cordis-h2020projects.csv</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitConfig"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE isStart="true" name="setProgrammeFileURL" type="SetEnvParameter">
|
|
<DESCRIPTION>sets the URL to download the programme file</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">programmeFileURL</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">https://cordis.europa.eu/data/reference/cordisref-h2020programmes.csv</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitConfig"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE isStart="true" name="setTopicFileURL" type="SetEnvParameter">
|
|
<DESCRIPTION>sets the URL to download the topics file</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">topicFileURL</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">https://cordis.europa.eu/data/reference/cordisref-h2020topics.xlsx</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitConfig"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE isStart="true" name="setSheetName" type="SetEnvParameter">
|
|
<DESCRIPTION>sets the name of the sheet in the topic file to be read</DESCRIPTION>
|
|
<PARAMETERS>
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">sheetName</PARAM>
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">Topics</PARAM>
|
|
</PARAMETERS>
|
|
<ARCS>
|
|
<ARC to="waitConfig"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE isJoin="true" name="waitConfig">
|
|
<DESCRIPTION>wait configurations</DESCRIPTION>
|
|
<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' : 'h2020classification',
|
|
'jobProperty' : 'export_action_set_h2020classification',
|
|
'enablingProperty' : 'active_h2020classification',
|
|
'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="h2020ClassificationUpdate"/>
|
|
</ARCS>
|
|
</NODE>
|
|
<NODE name="h2020ClassificationUpdate" type="SubmitHadoopJob">
|
|
<DESCRIPTION>prepare updates for the H2020 Classification</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">
|
|
{
|
|
'outputPath': 'outputPath',
|
|
'sheetName':'sheetName',
|
|
'projectFileURL' : 'projectFileURL',
|
|
'programmeFileURL' : 'programmeFileURL',
|
|
'topicFileURL':'topicFileURL'
|
|
}
|
|
</PARAM>
|
|
<PARAM managedBy="system" name="params" required="true" type="string">
|
|
{
|
|
'oozie.wf.application.path' : '/lib/dnet/PROD/actionmanager/project/oozie_app',
|
|
'workingDir' : '/tmp/prod_provision/working_dir/h2020classification',
|
|
'postgresURL':'',
|
|
'postgresUser':'',
|
|
'postgresPassword':''
|
|
}
|
|
</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_20210524_084803_740</LAST_EXECUTION_ID>
|
|
<LAST_EXECUTION_DATE>2021-05-24T09:05:50+00:00</LAST_EXECUTION_DATE>
|
|
<LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
|
|
<LAST_EXECUTION_ERROR/>
|
|
</STATUS>
|
|
</BODY>
|
|
</RESOURCE_PROFILE> |