|
1
|
<RESOURCE_PROFILE xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
|
|
2
|
<HEADER>
|
|
3
|
<RESOURCE_IDENTIFIER value="66fbd899-a981-4fd7-b27b-e546f8823ac7_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
|
|
4
|
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
|
|
5
|
<RESOURCE_KIND value="WorkflowDSResources"/>
|
|
6
|
<RESOURCE_URI value=""/>
|
|
7
|
<DATE_OF_CREATION value="2018-10-31T11:02:13+00:00"/>
|
|
8
|
</HEADER>
|
|
9
|
<BODY>
|
|
10
|
<WORKFLOW_NAME>Softwares Exporter</WORKFLOW_NAME>
|
|
11
|
<WORKFLOW_TYPE>Export Infospace</WORKFLOW_TYPE>
|
|
12
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
|
13
|
<CONFIGURATION start="manual">
|
|
14
|
<NODE isStart="true" name="setOutputPath" type="SetEnvParameter">
|
|
15
|
<DESCRIPTION>Set the output file path on HDFS</DESCRIPTION>
|
|
16
|
<PARAMETERS>
|
|
17
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">outputPath</PARAM>
|
|
18
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/organizations.2018.10.31</PARAM>
|
|
19
|
</PARAMETERS>
|
|
20
|
<ARCS>
|
|
21
|
<ARC to="cleanupOutput"/>
|
|
22
|
</ARCS>
|
|
23
|
</NODE>
|
|
24
|
<NODE isStart="true" name="hadoopConfig" type="SetClusterAndTable">
|
|
25
|
<DESCRIPTION>Set table name</DESCRIPTION>
|
|
26
|
<PARAMETERS>
|
|
27
|
<PARAM managedBy="system" name="cluster" required="true" type="string">DM</PARAM>
|
|
28
|
<PARAM managedBy="system" name="tableParam" required="true" type="string">tableName</PARAM>
|
|
29
|
<PARAM managedBy="user" name="table" required="true" type="string">db_openaireplus_services_beta</PARAM>
|
|
30
|
</PARAMETERS>
|
|
31
|
<ARCS>
|
|
32
|
<ARC to="cleanupOutput"/>
|
|
33
|
</ARCS>
|
|
34
|
</NODE>
|
|
35
|
<NODE isJoin="true" name="cleanupOutput" type="DeleteHdfsPathJob">
|
|
36
|
<DESCRIPTION>CSV files cleanup</DESCRIPTION>
|
|
37
|
<PARAMETERS>
|
|
38
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
|
39
|
{
|
|
40
|
'path' : 'outputPath',
|
|
41
|
'cluster' : 'cluster'
|
|
42
|
}
|
|
43
|
</PARAM>
|
|
44
|
</PARAMETERS>
|
|
45
|
<ARCS>
|
|
46
|
<ARC to="export"/>
|
|
47
|
</ARCS>
|
|
48
|
</NODE>
|
|
49
|
<NODE name="export" type="SubmitHadoopJob">
|
|
50
|
<DESCRIPTION>export the information space as sequence file</DESCRIPTION>
|
|
51
|
<PARAMETERS>
|
|
52
|
<PARAM managedBy="system" name="hadoopJob" required="true" type="string">informationSpaceSoftwareExportJob</PARAM>
|
|
53
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
|
54
|
{
|
|
55
|
'cluster' : 'cluster',
|
|
56
|
'hbase.mapred.inputtable' : 'tableName',
|
|
57
|
'hbase.mapreduce.inputtable' : 'tableName',
|
|
58
|
'mapred.output.dir' : 'outputPath'
|
|
59
|
}
|
|
60
|
</PARAM>
|
|
61
|
</PARAMETERS>
|
|
62
|
<ARCS>
|
|
63
|
<ARC to="success"/>
|
|
64
|
</ARCS>
|
|
65
|
</NODE>
|
|
66
|
</CONFIGURATION>
|
|
67
|
<STATUS>
|
|
68
|
<LAST_EXECUTION_ID>wf_20181031_114608_13</LAST_EXECUTION_ID>
|
|
69
|
<LAST_EXECUTION_DATE>2018-10-31T11:49:33+00:00</LAST_EXECUTION_DATE>
|
|
70
|
<LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
|
|
71
|
<LAST_EXECUTION_ERROR/>
|
|
72
|
</STATUS>
|
|
73
|
</BODY>
|
|
74
|
</RESOURCE_PROFILE>
|