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="374d07a1-2a76-470a-ab54-a7d6c35eab9b_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>Deduplication</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="validValues(['result', 'organization', 'person'])" required="true" type="string" name="entityType" managedBy="user"></PARAM>
|
30
|
<PARAM required="true" type="string" name="dedupConfigSequence" managedBy="user"></PARAM>
|
31
|
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM>
|
32
|
</PARAMETERS>
|
33
|
<ARCS>
|
34
|
<ARC to="cleanupCsv" />
|
35
|
</ARCS>
|
36
|
</NODE>
|
37
|
|
38
|
<NODE name="hadoopConfig" type="SetClusterAndTable" isStart="true">
|
39
|
<DESCRIPTION>Set table name</DESCRIPTION>
|
40
|
<PARAMETERS>
|
41
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
42
|
<PARAM required="true" type="string" name="table" managedBy="user">db_stdl</PARAM>
|
43
|
<PARAM required="true" type="string" name="tableParam" managedBy="system">tableName</PARAM>
|
44
|
</PARAMETERS>
|
45
|
<ARCS>
|
46
|
<ARC to="cleanupCsv" />
|
47
|
</ARCS>
|
48
|
</NODE>
|
49
|
|
50
|
<NODE name="cleanupCsv" type="DeleteHdfsPathJob">
|
51
|
<DESCRIPTION>CSV files cleanup</DESCRIPTION>
|
52
|
<PARAMETERS>
|
53
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
54
|
{
|
55
|
'path' : 'csvPath',
|
56
|
'cluster' : 'cluster'
|
57
|
}
|
58
|
</PARAM>
|
59
|
</PARAMETERS>
|
60
|
<ARCS>
|
61
|
<ARC to="roots2CSV" />
|
62
|
</ARCS>
|
63
|
</NODE>
|
64
|
|
65
|
<NODE name="roots2CSV" type="SubmitHadoopJob">
|
66
|
<DESCRIPTION>export the representative entities as CSV files</DESCRIPTION>
|
67
|
<PARAMETERS>
|
68
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupRootsToCSVJob</PARAM>
|
69
|
<PARAM required="true" type="string" name="sysParams" managedBy="system">
|
70
|
{
|
71
|
'hbase.mapred.inputtable' : 'hbase.mapred.datatable',
|
72
|
'hbase.mapreduce.inputtable' : 'hbase.mapred.datatable'
|
73
|
}
|
74
|
</PARAM>
|
75
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
76
|
{
|
77
|
'cluster' : 'cluster',
|
78
|
'dedup.wf.conf' : 'dedup.wf.conf',
|
79
|
'relClasses' : 'relClasses',
|
80
|
'entityTypeId' : 'entityTypeId',
|
81
|
'entityType' : 'entityType',
|
82
|
'hbase.mapred.inputtable' : 'tableName',
|
83
|
'hbase.mapreduce.inputtable' : 'tableName',
|
84
|
'mapred.output.dir' : 'csvPath'
|
85
|
}
|
86
|
</PARAM>
|
87
|
<PARAM required="true" type="string" name="params" managedBy="system">
|
88
|
{
|
89
|
'entityTypeId' : '50',
|
90
|
'entityType' : 'result'
|
91
|
}
|
92
|
</PARAM>
|
93
|
</PARAMETERS>
|
94
|
<ARCS>
|
95
|
<ARC to="success" />
|
96
|
</ARCS>
|
97
|
</NODE>
|
98
|
</CONFIGURATION>
|
99
|
<STATUS />
|
100
|
</BODY>
|
101
|
</RESOURCE_PROFILE>
|
102
|
|
103
|
|
104
|
|
105
|
|
106
|
|
107
|
|