1
|
<RESOURCE_PROFILE xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
|
2
|
<HEADER>
|
3
|
<RESOURCE_IDENTIFIER value="cc6cf74c-5fef-4cdf-9f93-bd02b70709d2_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
|
4
|
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
|
5
|
<RESOURCE_KIND value="WorkflowDSResources"/>
|
6
|
<RESOURCE_URI value=""/>
|
7
|
<DATE_OF_CREATION value="2006-05-04T18:13:51.0Z"/>
|
8
|
</HEADER>
|
9
|
<BODY>
|
10
|
<WORKFLOW_NAME>Dedup persons</WORKFLOW_NAME>
|
11
|
<WORKFLOW_TYPE>Deduplication</WORKFLOW_TYPE>
|
12
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
13
|
<CONFIGURATION start="manual">
|
14
|
<NODE isStart="true" name="fetchRelClasses" type="FetchRelClasses">
|
15
|
<DESCRIPTION/>
|
16
|
<PARAMETERS>
|
17
|
<PARAM managedBy="system" name="relClassesProperty" required="true" type="string">dnet.openaire.model.relclasses.xquery</PARAM>
|
18
|
<PARAM managedBy="system" name="relClassesName" required="true" type="string">relClasses</PARAM>
|
19
|
</PARAMETERS>
|
20
|
<ARCS>
|
21
|
<ARC to="waitConf"/>
|
22
|
</ARCS>
|
23
|
</NODE>
|
24
|
<NODE name="hadoopConfig" type="SetClusterAndTable" isStart="true">
|
25
|
<DESCRIPTION>Set table name</DESCRIPTION>
|
26
|
<PARAMETERS>
|
27
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
28
|
<PARAM required="true" type="string" name="table" managedBy="user">db_person_gt</PARAM>
|
29
|
<PARAM required="true" type="string" name="tableParam" managedBy="system">tableName</PARAM>
|
30
|
</PARAMETERS>
|
31
|
<ARCS>
|
32
|
<ARC to="waitConf" />
|
33
|
</ARCS>
|
34
|
</NODE>
|
35
|
|
36
|
<NODE name="setDedupConfigs" type="SetDedupConfiguration" isStart="true">
|
37
|
<DESCRIPTION>Set Dedup conf</DESCRIPTION>
|
38
|
<PARAMETERS>
|
39
|
<PARAM required="true" type="string" name="dedupConfigSequence" managedBy="user">dedup-similarity-person</PARAM>
|
40
|
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM>
|
41
|
</PARAMETERS>
|
42
|
<ARCS>
|
43
|
<ARC to="waitConf" />
|
44
|
</ARCS>
|
45
|
</NODE>
|
46
|
|
47
|
<NODE name="waitConf" isJoin="true">
|
48
|
<DESCRIPTION/>
|
49
|
<PARAMETERS/>
|
50
|
<ARCS>
|
51
|
<ARC to="dedupPhase" />
|
52
|
</ARCS>
|
53
|
</NODE>
|
54
|
|
55
|
<NODE name="dedupPhase" type="DedupConfigurationAwareJob">
|
56
|
<DESCRIPTION>Dedup M/R job</DESCRIPTION>
|
57
|
<PARAMETERS>
|
58
|
<PARAM managedBy="system" name="hadoopJob" required="true" type="string">dedupPersonJob</PARAM>
|
59
|
<PARAM managedBy="system" name="dedupConfigSequenceParam" type="string" required="true">dedup.conf.queue</PARAM>
|
60
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
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="buildRoots"/>
|
74
|
</ARCS>
|
75
|
</NODE>
|
76
|
<NODE name="buildRoots" type="SubmitHadoopJob">
|
77
|
<DESCRIPTION>redirect rels</DESCRIPTION>
|
78
|
<PARAMETERS>
|
79
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupBuildRootsJob</PARAM>
|
80
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
81
|
{
|
82
|
'dedup.conf' : 'dedup.conf',
|
83
|
'relClasses' : 'relClasses',
|
84
|
'entityTypeId' : 'entityTypeId',
|
85
|
'entityType' : 'entityType',
|
86
|
'cluster' : 'cluster',
|
87
|
'hbase.mapred.inputtable' : 'tableName',
|
88
|
'hbase.mapreduce.inputtable' : 'tableName',
|
89
|
'hbase.mapred.outputtable' : 'tableName',
|
90
|
'hbase.mapreduce.outputtable' : 'tableName'
|
91
|
}
|
92
|
</PARAM>
|
93
|
</PARAMETERS>
|
94
|
<ARCS>
|
95
|
<ARC to="success"/>
|
96
|
</ARCS>
|
97
|
</NODE>
|
98
|
</CONFIGURATION>
|
99
|
<STATUS>
|
100
|
<LAST_EXECUTION_ID>wf_20150513_160956_878</LAST_EXECUTION_ID>
|
101
|
<LAST_EXECUTION_DATE>2015-05-13T16:58:47+00:00</LAST_EXECUTION_DATE>
|
102
|
<LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
|
103
|
<LAST_EXECUTION_ERROR/>
|
104
|
</STATUS>
|
105
|
</BODY>
|
106
|
</RESOURCE_PROFILE>
|