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="ffaf21de-62f6-43b6-93d6-de0c33a38e28_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>Offline Deduplication</WORKFLOW_NAME>
|
12
|
<WORKFLOW_TYPE>Deduplication</WORKFLOW_TYPE>
|
13
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
14
|
<CONFIGURATION start="manual">
|
15
|
<NODE name="fetchRelClasses" type="FetchRelClasses" isStart="true">
|
16
|
<DESCRIPTION />
|
17
|
<PARAMETERS>
|
18
|
<PARAM name="relClassesProperty" type="string" managedBy="system" required="true">dnet.openaire.model.relclasses.xquery</PARAM>
|
19
|
<PARAM name="relClassesName" type="string" managedBy="system" required="true">relClasses</PARAM>
|
20
|
</PARAMETERS>
|
21
|
<ARCS>
|
22
|
<ARC to="checkConf" />
|
23
|
</ARCS>
|
24
|
</NODE>
|
25
|
<NODE name="setDedupConfigs" type="SetDedupConfiguration" isStart="true">
|
26
|
<DESCRIPTION>Set Dedup conf</DESCRIPTION>
|
27
|
<PARAMETERS>
|
28
|
<PARAM function="obtainValues('dedupOrchestrations', {})" required="true" type="string" name="dedupConfigSequence" managedBy="user"></PARAM>
|
29
|
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM>
|
30
|
</PARAMETERS>
|
31
|
<ARCS>
|
32
|
<ARC to="checkConf" />
|
33
|
</ARCS>
|
34
|
</NODE>
|
35
|
<NODE name="hadoopConfig" type="SetClusterAndTable" isStart="true">
|
36
|
<DESCRIPTION>Set table name</DESCRIPTION>
|
37
|
<PARAMETERS>
|
38
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
39
|
<PARAM required="true" type="string" name="tableParam" managedBy="system">tableName</PARAM>
|
40
|
</PARAMETERS>
|
41
|
<ARCS>
|
42
|
<ARC to="checkConf" />
|
43
|
</ARCS>
|
44
|
</NODE>
|
45
|
<NODE name="checkConf" type="DedupCheckConfiguration" isJoin="true">
|
46
|
<DESCRIPTION/>
|
47
|
<PARAMETERS>
|
48
|
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM>
|
49
|
</PARAMETERS>
|
50
|
<ARCS>
|
51
|
<ARC to="deduplicateScan" />
|
52
|
</ARCS>
|
53
|
</NODE>
|
54
|
|
55
|
<NODE name="deduplicateScan" type="DuplicateScanJob">
|
56
|
<DESCRIPTION>Dup Scan</DESCRIPTION>
|
57
|
<PARAMETERS>
|
58
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupCandidateScanJob</PARAM>
|
59
|
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM>
|
60
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
61
|
{
|
62
|
'entityTypeId' : 'entityTypeId',
|
63
|
'entityType' : 'entityType',
|
64
|
'cluster' : 'cluster',
|
65
|
'hbase.mapred.inputtable' : 'tableName',
|
66
|
'hbase.mapred.outputtable' : 'tableName',
|
67
|
'hbase.mapreduce.inputtable' : 'tableName',
|
68
|
'hbase.mapreduce.outputtable' : 'tableName'
|
69
|
}
|
70
|
</PARAM>
|
71
|
</PARAMETERS>
|
72
|
<ARCS>
|
73
|
<ARC to="deduplicateScan" />
|
74
|
<ARC name="done" to="dedupGrouper" />
|
75
|
</ARCS>
|
76
|
</NODE>
|
77
|
|
78
|
<NODE name="dedupGrouper" type="DedupGrouperJob">
|
79
|
<DESCRIPTION>dedup grouper</DESCRIPTION>
|
80
|
<PARAMETERS>
|
81
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupGrouperJob</PARAM>
|
82
|
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM>
|
83
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
84
|
{
|
85
|
'dedup.conf' : 'dedup.conf',
|
86
|
'entityTypeId' : 'entityTypeId',
|
87
|
'entityType' : 'entityType',
|
88
|
'cluster' : 'cluster',
|
89
|
'hbase.mapred.inputtable' : 'tableName',
|
90
|
'hbase.mapred.outputtable' : 'tableName',
|
91
|
'hbase.mapreduce.inputtable' : 'tableName',
|
92
|
'hbase.mapreduce.outputtable' : 'tableName'
|
93
|
}
|
94
|
</PARAM>
|
95
|
</PARAMETERS>
|
96
|
<ARCS>
|
97
|
<ARC to="dedupGrouper" />
|
98
|
<ARC name="done" to="findRoots" />
|
99
|
</ARCS>
|
100
|
</NODE>
|
101
|
<NODE name="findRoots" type="SubmitHadoopJob">
|
102
|
<DESCRIPTION>find roots</DESCRIPTION>
|
103
|
<PARAMETERS>
|
104
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupFindPersonRootsJob</PARAM>
|
105
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
106
|
{
|
107
|
'dedup.conf' : 'dedup.conf',
|
108
|
'entityTypeId' : 'entityTypeId',
|
109
|
'entityType' : 'entityType',
|
110
|
'cluster' : 'cluster',
|
111
|
'hbase.mapred.inputtable' : 'tableName',
|
112
|
'hbase.mapred.outputtable' : 'tableName',
|
113
|
'hbase.mapreduce.inputtable' : 'tableName',
|
114
|
'hbase.mapreduce.outputtable' : 'tableName'
|
115
|
}
|
116
|
</PARAM>
|
117
|
</PARAMETERS>
|
118
|
<ARCS>
|
119
|
<ARC to="deleteSimRels" />
|
120
|
</ARCS>
|
121
|
</NODE>
|
122
|
<NODE name="deleteSimRels" type="SubmitHadoopJob">
|
123
|
<DESCRIPTION>redirect rels</DESCRIPTION>
|
124
|
<PARAMETERS>
|
125
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">deleteSimRelJob</PARAM>
|
126
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
127
|
{
|
128
|
'dedup.conf' : 'dedup.conf',
|
129
|
'entityTypeId' : 'entityTypeId',
|
130
|
'entityType' : 'entityType',
|
131
|
'cluster' : 'cluster',
|
132
|
'hbase.mapred.inputtable' : 'tableName',
|
133
|
'hbase.mapreduce.inputtable' : 'tableName',
|
134
|
'hbase.mapred.outputtable' : 'tableName',
|
135
|
'hbase.mapreduce.outputtable' : 'tableName'
|
136
|
}
|
137
|
</PARAM>
|
138
|
</PARAMETERS>
|
139
|
<ARCS>
|
140
|
<ARC to="success" />
|
141
|
</ARCS>
|
142
|
</NODE>
|
143
|
</CONFIGURATION>
|
144
|
<STATUS />
|
145
|
</BODY>
|
146
|
</RESOURCE_PROFILE>
|