dnet-hadoop/dhp-workflows/dhp-worfklow-profiles/src/main/resources/eu/dnetlib/dhp/provision/03_graph2hive.xml

74 lines
3.6 KiB
XML

<RESOURCE_PROFILE>
<HEADER>
<RESOURCE_IDENTIFIER value="5d750977-bec2-47f4-97bb-1b7500e4704e_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
<RESOURCE_KIND value="WorkflowDSResources"/>
<RESOURCE_URI value=""/>
<DATE_OF_CREATION value="2021-07-28T07:49:37+00:00"/>
</HEADER>
<BODY>
<WORKFLOW_NAME>Graph to HiveDB [PROD]</WORKFLOW_NAME>
<WORKFLOW_TYPE>Data Provision</WORKFLOW_TYPE>
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
<CONFIGURATION start="manual">
<NODE isStart="true" name="setInputPath" type="SetEnvParameter">
<DESCRIPTION>Set the path containing the AGGREGATOR graph</DESCRIPTION>
<PARAMETERS>
<PARAM managedBy="system" name="parameterName" required="true" type="string">inputPath</PARAM>
<PARAM managedBy="user" name="parameterValue" required="true" type="string"></PARAM>
</PARAMETERS>
<ARCS>
<ARC to="waitConfig"/>
</ARCS>
</NODE>
<NODE isStart="true" name="setHiveDbName" type="SetEnvParameter">
<DESCRIPTION>Set the target path to store the RAW graph</DESCRIPTION>
<PARAMETERS>
<PARAM managedBy="system" name="parameterName" required="true" type="string">hiveDbName</PARAM>
<PARAM managedBy="user" name="parameterValue" required="true" type="string"></PARAM>
</PARAMETERS>
<ARCS>
<ARC to="waitConfig"/>
</ARCS>
</NODE>
<NODE isJoin="true" name="waitConfig">
<DESCRIPTION>wait configurations</DESCRIPTION>
<PARAMETERS/>
<ARCS>
<ARC to="graph2hive"/>
</ARCS>
</NODE>
<NODE name="graph2hive" type="SubmitHadoopJob">
<DESCRIPTION>create the AGGREGATOR graph</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">
{
'inputPath' : 'inputPath',
'hiveDbName' : 'hiveDbName'
}
</PARAM>
<PARAM managedBy="system" name="params" required="true" type="string">
{
'oozie.wf.application.path' : '/lib/dnet/PROD/oa/graph/hive/oozie_app',
'sparkDriverMemory' : '4G',
'sparkExecutorMemory' : '10G',
'sparkExecutorCores' : '3'
}
</PARAM>
<PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
</PARAMETERS>
<ARCS>
<ARC to="success"/>
</ARCS>
</NODE>
</CONFIGURATION>
<STATUS>
<LAST_EXECUTION_ID>wf_20210728_075001_400</LAST_EXECUTION_ID>
<LAST_EXECUTION_DATE>2021-07-28T08:04:00+00:00</LAST_EXECUTION_DATE>
<LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
<LAST_EXECUTION_ERROR/>
</STATUS>
</BODY>
</RESOURCE_PROFILE>