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="f6a4c2e5-a663-4700-844f-1b753484efee_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>Dedup organizations</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="dedupPhase" />
|
23
|
</ARCS>
|
24
|
</NODE>
|
25
|
<NODE name="configure" type="LoadDedupConfiguration" isStart="true">
|
26
|
<DESCRIPTION>Load Dedup conf</DESCRIPTION>
|
27
|
<PARAMETERS>
|
28
|
<PARAM required="true" type="string" name="entityType" managedBy="system">organization</PARAM>
|
29
|
</PARAMETERS>
|
30
|
<ARCS>
|
31
|
<ARC to="dedupPhase" />
|
32
|
</ARCS>
|
33
|
</NODE>
|
34
|
<NODE name="dedupPhase" type="SubmitHadoopJob" isJoin="true">
|
35
|
<DESCRIPTION>Dedup M/R job</DESCRIPTION>
|
36
|
<PARAMETERS>
|
37
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupCandidateScanJob</PARAM>
|
38
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
39
|
<PARAM required="true" type="string" name="sysParams" managedBy="system">
|
40
|
{
|
41
|
'hbase.mapred.inputtable' : 'hbase.mapred.datatable',
|
42
|
'hbase.mapred.outputtable' : 'hbase.mapred.datatable',
|
43
|
'hbase.mapreduce.inputtable' : 'hbase.mapred.datatable',
|
44
|
'hbase.mapreduce.outputtable' : 'hbase.mapred.datatable'
|
45
|
}
|
46
|
</PARAM>
|
47
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
48
|
{
|
49
|
'dedup.pace.conf' : 'dedup.pace.conf',
|
50
|
'dedup.wf.conf' : 'dedup.wf.conf'
|
51
|
}
|
52
|
</PARAM>
|
53
|
<PARAM required="true" type="string" name="params" managedBy="system">
|
54
|
{
|
55
|
'entityTypeId' : '20',
|
56
|
'entityType' : 'organization'
|
57
|
}
|
58
|
</PARAM>
|
59
|
</PARAMETERS>
|
60
|
<ARCS>
|
61
|
<ARC to="dedupGrouper" />
|
62
|
</ARCS>
|
63
|
</NODE>
|
64
|
<NODE name="dedupGrouper" type="DedupGrouperJob">
|
65
|
<DESCRIPTION>dedup grouper</DESCRIPTION>
|
66
|
<PARAMETERS>
|
67
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupGrouperJob</PARAM>
|
68
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
69
|
<PARAM required="true" type="string" name="sysParams" managedBy="system">
|
70
|
{
|
71
|
'hbase.mapred.inputtable' : 'hbase.mapred.datatable',
|
72
|
'hbase.mapred.outputtable' : 'hbase.mapred.datatable',
|
73
|
'hbase.mapreduce.inputtable' : 'hbase.mapred.datatable',
|
74
|
'hbase.mapreduce.outputtable' : 'hbase.mapred.datatable'
|
75
|
}
|
76
|
</PARAM>
|
77
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
78
|
{
|
79
|
'dedup.pace.conf' : 'dedup.pace.conf',
|
80
|
'dedup.wf.conf' : 'dedup.wf.conf'
|
81
|
}
|
82
|
</PARAM>
|
83
|
<PARAM required="true" type="string" name="params" managedBy="system">
|
84
|
{
|
85
|
'entityTypeId' : '20'
|
86
|
}
|
87
|
</PARAM>
|
88
|
</PARAMETERS>
|
89
|
<ARCS>
|
90
|
<ARC to="dedupGrouper" />
|
91
|
<ARC name="done" to="findRoots" />
|
92
|
</ARCS>
|
93
|
</NODE>
|
94
|
|
95
|
<NODE name="findRoots" type="SubmitHadoopJob">
|
96
|
<DESCRIPTION>find roots</DESCRIPTION>
|
97
|
<PARAMETERS>
|
98
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupFindRootsJob</PARAM>
|
99
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
100
|
<PARAM required="true" type="string" name="sysParams" managedBy="system">
|
101
|
{
|
102
|
'hbase.mapred.inputtable' : 'hbase.mapred.datatable',
|
103
|
'hbase.mapred.outputtable' : 'hbase.mapred.datatable',
|
104
|
'hbase.mapreduce.inputtable' : 'hbase.mapred.datatable',
|
105
|
'hbase.mapreduce.outputtable' : 'hbase.mapred.datatable'
|
106
|
}
|
107
|
</PARAM>
|
108
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
109
|
{
|
110
|
'dedup.pace.conf' : 'dedup.pace.conf',
|
111
|
'dedup.wf.conf' : 'dedup.wf.conf'
|
112
|
}
|
113
|
</PARAM>
|
114
|
<PARAM required="true" type="string" name="params" managedBy="system">
|
115
|
{
|
116
|
'entityTypeId' : '20',
|
117
|
'entityType' : 'organization'
|
118
|
}
|
119
|
</PARAM>
|
120
|
</PARAMETERS>
|
121
|
<ARCS>
|
122
|
<ARC to="buildRoots" />
|
123
|
</ARCS>
|
124
|
</NODE>
|
125
|
<NODE name="buildRoots" type="SubmitHadoopJob">
|
126
|
<DESCRIPTION>redirect rels</DESCRIPTION>
|
127
|
<PARAMETERS>
|
128
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupBuildRootsJob</PARAM>
|
129
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
130
|
<PARAM required="true" type="string" name="sysParams" managedBy="system">
|
131
|
{
|
132
|
'hbase.mapred.inputtable' : 'hbase.mapred.datatable',
|
133
|
'hbase.mapred.outputtable' : 'hbase.mapred.datatable',
|
134
|
'hbase.mapreduce.inputtable' : 'hbase.mapred.datatable',
|
135
|
'hbase.mapreduce.outputtable' : 'hbase.mapred.datatable'
|
136
|
}
|
137
|
</PARAM>
|
138
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
139
|
{
|
140
|
'dedup.pace.conf' : 'dedup.pace.conf',
|
141
|
'dedup.wf.conf' : 'dedup.wf.conf',
|
142
|
'relClasses' : 'relClasses'
|
143
|
}
|
144
|
</PARAM>
|
145
|
<PARAM required="true" type="string" name="params" managedBy="system">
|
146
|
{
|
147
|
'entityTypeId' : '20'
|
148
|
}
|
149
|
</PARAM>
|
150
|
</PARAMETERS>
|
151
|
<ARCS>
|
152
|
<ARC to="success" />
|
153
|
</ARCS>
|
154
|
</NODE>
|
155
|
</CONFIGURATION>
|
156
|
<STATUS />
|
157
|
</BODY>
|
158
|
</RESOURCE_PROFILE>
|