|
1
|
<RESOURCE_PROFILE>
|
|
2
|
<HEADER>
|
|
3
|
<RESOURCE_IDENTIFIER value="b030513b-69af-4a9f-90c7-90b96bd5eb5c_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
|
|
4
|
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
|
|
5
|
<RESOURCE_KIND value="WorkflowDSResources"/>
|
|
6
|
<RESOURCE_URI value=""/>
|
|
7
|
<DATE_OF_CREATION value="2018-11-06T11:27:07+00:00"/>
|
|
8
|
</HEADER>
|
|
9
|
<BODY>
|
|
10
|
<WORKFLOW_NAME>XML records counts</WORKFLOW_NAME>
|
|
11
|
<WORKFLOW_TYPE>Data Provision</WORKFLOW_TYPE>
|
|
12
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
|
13
|
<CONFIGURATION start="manual"><!-- PREPARE NODES -->
|
|
14
|
<NODE isStart="true" name="setInputFilesPath" type="SetEnvParameter">
|
|
15
|
<DESCRIPTION>set the input path on hdfs to read xml records</DESCRIPTION>
|
|
16
|
<PARAMETERS>
|
|
17
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">inputHdfsPath</PARAM>
|
|
18
|
<PARAM managedBy="user" name="parameterValue" required="false" type="string">/tmp/indexrecords_db_openaireplus_services_TMF.seq</PARAM>
|
|
19
|
</PARAMETERS>
|
|
20
|
<ARCS>
|
|
21
|
<ARC to="countXML"/>
|
|
22
|
</ARCS>
|
|
23
|
</NODE>
|
|
24
|
<NODE isJoin="true" name="countXML" type="SubmitHadoopJob">
|
|
25
|
<DESCRIPTION>Run M/R count Job</DESCRIPTION>
|
|
26
|
<PARAMETERS>
|
|
27
|
<PARAM managedBy="system" name="cluster" required="true" type="string">DM</PARAM>
|
|
28
|
<PARAM managedBy="system" name="hadoopJob" required="true" type="string">xmlRecordCounterJob</PARAM>
|
|
29
|
<PARAM managedBy="user" name="simulation" required="true" type="boolean">false</PARAM>
|
|
30
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
|
31
|
{
|
|
32
|
'mapred.input.dir' : 'inputHdfsPath'
|
|
33
|
}
|
|
34
|
</PARAM>
|
|
35
|
</PARAMETERS>
|
|
36
|
<ARCS>
|
|
37
|
<ARC to="success"/>
|
|
38
|
</ARCS>
|
|
39
|
</NODE>
|
|
40
|
</CONFIGURATION>
|
|
41
|
<STATUS>
|
|
42
|
<LAST_EXECUTION_ID>wf_20181022_103455_527</LAST_EXECUTION_ID>
|
|
43
|
<LAST_EXECUTION_DATE>2018-10-22T12:44:29+00:00</LAST_EXECUTION_DATE>
|
|
44
|
<LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
|
|
45
|
<LAST_EXECUTION_ERROR/>
|
|
46
|
</STATUS>
|
|
47
|
</BODY>
|
|
48
|
</RESOURCE_PROFILE>
|