|
1
|
<?xml version="1.0" encoding="UTF-8"?>
|
|
2
|
<RESOURCE_PROFILE xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
|
|
3
|
<HEADER>
|
|
4
|
<RESOURCE_IDENTIFIER value="7d183ac5-5aae-4022-84b6-37a9e4f69ab0_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl" />
|
|
5
|
<RESOURCE_TYPE value="WorkflowDSResourceType" />
|
|
6
|
<RESOURCE_KIND value="WorkflowDSResources" />
|
|
7
|
<RESOURCE_URI value="" />
|
|
8
|
<DATE_OF_CREATION value="2006-05-04T18:13:51.0Z" />
|
|
9
|
</HEADER>
|
|
10
|
<BODY>
|
|
11
|
<WORKFLOW_NAME>Deduplication export</WORKFLOW_NAME>
|
|
12
|
<WORKFLOW_TYPE>Export to HDFS</WORKFLOW_TYPE>
|
|
13
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
|
14
|
<CONFIGURATION start="manual">
|
|
15
|
<NODE name="setCsvPath" type="SetEnvParameter" isStart="true">
|
|
16
|
<DESCRIPTION>Set the CSV file path on HDFS</DESCRIPTION>
|
|
17
|
<PARAMETERS>
|
|
18
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">csvPath</PARAM>
|
|
19
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string"></PARAM>
|
|
20
|
</PARAMETERS>
|
|
21
|
<ARCS>
|
|
22
|
<ARC to="cleanupCsv" />
|
|
23
|
</ARCS>
|
|
24
|
</NODE>
|
|
25
|
|
|
26
|
<NODE name="setDedupConfigs" type="SetDedupConfiguration" isStart="true">
|
|
27
|
<DESCRIPTION>Set Dedup conf</DESCRIPTION>
|
|
28
|
<PARAMETERS>
|
|
29
|
<PARAM function="obtainValues('dedupOrchestrations', {})" required="true" type="string" name="dedupConfigSequence" managedBy="user"></PARAM>
|
|
30
|
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM>
|
|
31
|
</PARAMETERS>
|
|
32
|
<ARCS>
|
|
33
|
<ARC to="cleanupCsv" />
|
|
34
|
</ARCS>
|
|
35
|
</NODE>
|
|
36
|
|
|
37
|
<NODE name="hadoopConfig" type="SetClusterAndTable" isStart="true">
|
|
38
|
<DESCRIPTION>Set table name</DESCRIPTION>
|
|
39
|
<PARAMETERS>
|
|
40
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
|
41
|
<PARAM required="true" type="string" name="tableParam" managedBy="system">tableName</PARAM>
|
|
42
|
<PARAM required="true" type="string" name="table" managedBy="user"/>
|
|
43
|
</PARAMETERS>
|
|
44
|
<ARCS>
|
|
45
|
<ARC to="cleanupCsv" />
|
|
46
|
</ARCS>
|
|
47
|
</NODE>
|
|
48
|
|
|
49
|
<NODE name="cleanupCsv" type="DeleteHdfsPathJob" isJoin="true">
|
|
50
|
<DESCRIPTION>CSV files cleanup</DESCRIPTION>
|
|
51
|
<PARAMETERS>
|
|
52
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
|
53
|
{
|
|
54
|
'path' : 'csvPath',
|
|
55
|
'cluster' : 'cluster'
|
|
56
|
}
|
|
57
|
</PARAM>
|
|
58
|
</PARAMETERS>
|
|
59
|
<ARCS>
|
|
60
|
<ARC to="export" />
|
|
61
|
</ARCS>
|
|
62
|
</NODE>
|
|
63
|
|
|
64
|
<NODE name="export" type="SubmitHadoopJob">
|
|
65
|
<DESCRIPTION>export the representative publications</DESCRIPTION>
|
|
66
|
<PARAMETERS>
|
|
67
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupRootsExportJob</PARAM>
|
|
68
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
|
69
|
{
|
|
70
|
'cluster' : 'cluster',
|
|
71
|
'entityTypeId' : 'entityTypeId',
|
|
72
|
'entityType' : 'entityType',
|
|
73
|
'hbase.mapred.inputtable' : 'tableName',
|
|
74
|
'hbase.mapreduce.inputtable' : 'tableName',
|
|
75
|
'mapred.output.dir' : 'csvPath'
|
|
76
|
}
|
|
77
|
</PARAM>
|
|
78
|
</PARAMETERS>
|
|
79
|
<ARCS>
|
|
80
|
<ARC to="success" />
|
|
81
|
</ARCS>
|
|
82
|
</NODE>
|
|
83
|
</CONFIGURATION>
|
|
84
|
<STATUS />
|
|
85
|
</BODY>
|
|
86
|
</RESOURCE_PROFILE>
|
|
87
|
|
|
88
|
|
|
89
|
|
|
90
|
|
|
91
|
|
|
92
|
|