|
1
|
<RESOURCE_PROFILE>
|
|
2
|
<HEADER>
|
|
3
|
<RESOURCE_IDENTIFIER value="01ed11e8-e874-4478-a8ac-83e63e9699e4_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ=="/>
|
|
4
|
<RESOURCE_TYPE value="WorkflowTemplateDSResourceType"/>
|
|
5
|
<RESOURCE_KIND value="WorkflowTemplateDSResources"/>
|
|
6
|
<RESOURCE_URI value=""/>
|
|
7
|
<DATE_OF_CREATION value="2016-06-25T10:24:32+00:00"/>
|
|
8
|
</HEADER>
|
|
9
|
<BODY>
|
|
10
|
<CONFIGURATION>
|
|
11
|
|
|
12
|
<PARAMETERS>
|
|
13
|
|
|
14
|
<PARAM name="hbaseTable" description="HBase table name" required="true" type="string"/>
|
|
15
|
<PARAM name="cluster" description="Hadoop cluster name" required="true" type="string"/>
|
|
16
|
|
|
17
|
<PARAM name="dedupConfigSequence" description="dedup configuration orchestration name" required="true" type="string"/>
|
|
18
|
<PARAM name="minDistWorkDir" description="work directory for the minDist algorithm" required="true" type="string"/>
|
|
19
|
|
|
20
|
<PARAM name="mdFormat" description="dedup index mdFormat" required="true" type="string" default="OPENAIRE"/>
|
|
21
|
<PARAM name="layout" description="dedup index layout" required="true" type="string" default="index"/>
|
|
22
|
<PARAM name="interpretation" description="dedup index interpretation" required="true" type="string" default="dedup"/>
|
|
23
|
|
|
24
|
</PARAMETERS>
|
|
25
|
|
|
26
|
<WORKFLOW>
|
|
27
|
|
|
28
|
<NODE isStart="true" name="setDedupConfigs" type="SetDedupConfiguration">
|
|
29
|
<DESCRIPTION>Set Dedup conf</DESCRIPTION>
|
|
30
|
<PARAMETERS>
|
|
31
|
<PARAM name="dedupConfigSequence" ref="dedupConfigSequence"/>
|
|
32
|
<PARAM name="dedupConfigSequenceParam" value="dedup.conf.queue"/>
|
|
33
|
<PARAM name="entityTypeNameParam" value="entityType"/>
|
|
34
|
<PARAM name="entityTypeIdParam" value="entityTypeId"/>
|
|
35
|
</PARAMETERS>
|
|
36
|
<ARCS>
|
|
37
|
<ARC to="resetPath"/>
|
|
38
|
</ARCS>
|
|
39
|
</NODE>
|
|
40
|
|
|
41
|
|
|
42
|
<NODE name="resetPath" type="CreateHdfsDirJob">
|
|
43
|
<DESCRIPTION>input files cleanup</DESCRIPTION>
|
|
44
|
<PARAMETERS>
|
|
45
|
<PARAM name="force" value="true"/>
|
|
46
|
<PARAM name="path" ref="minDistWorkDir"/>
|
|
47
|
<PARAM name="cluster" ref="cluster"/>
|
|
48
|
</PARAMETERS>
|
|
49
|
<ARCS>
|
|
50
|
<ARC to="deduplicateScan"/>
|
|
51
|
</ARCS>
|
|
52
|
</NODE>
|
|
53
|
|
|
54
|
|
|
55
|
<NODE name="deduplicateScan" type="DuplicateScanJob">
|
|
56
|
<DESCRIPTION>Dup Scan</DESCRIPTION>
|
|
57
|
<PARAMETERS>
|
|
58
|
<PARAM name="hadoopJob" value="dedupCandidateScanJob"/>
|
|
59
|
<PARAM name="dedupConfigurationOrchestration" env="dedup.conf.queue"/>
|
|
60
|
<PARAM name="jobParams">
|
|
61
|
<MAP>
|
|
62
|
<ENTRY key="entityTypeId" env="entityTypeId"/>
|
|
63
|
<ENTRY key="entityType" env="entityType"/>
|
|
64
|
<ENTRY key="cluster" ref="cluster"/>
|
|
65
|
<ENTRY key="hbase.mapred.inputtable" ref="hbaseTable"/>
|
|
66
|
<ENTRY key="hbase.mapred.outputtable" ref="hbaseTable"/>
|
|
67
|
<ENTRY key="hbase.mapreduce.inputtable" ref="hbaseTable"/>
|
|
68
|
<ENTRY key="hbase.mapreduce.outputtable" ref="hbaseTable"/>
|
|
69
|
</MAP>
|
|
70
|
</PARAM>
|
|
71
|
</PARAMETERS>
|
|
72
|
<ARCS>
|
|
73
|
<ARC to="deduplicateScan"/>
|
|
74
|
<ARC name="done" to="prepareActionSet"/>
|
|
75
|
<ARC name="no_dedup" to="doneActions"/>
|
|
76
|
</ARCS>
|
|
77
|
</NODE>
|
|
78
|
|
|
79
|
<NODE name="prepareActionSet" type="PrepareConfiguredActionSet">
|
|
80
|
<DESCRIPTION>prepare action sets</DESCRIPTION>
|
|
81
|
<PARAMETERS>
|
|
82
|
<PARAM name="dedupConfigSequence" ref="dedupConfigSequence"/>
|
|
83
|
<PARAM name="jobProperty" value="rawSetId"/>
|
|
84
|
<PARAM name="actionSetPathParam" value="actionSetPath"/>
|
|
85
|
<PARAM name="setsParam" value="sets"/>
|
|
86
|
</PARAMETERS>
|
|
87
|
<ARCS>
|
|
88
|
<ARC to="similarity2actions"/>
|
|
89
|
</ARCS>
|
|
90
|
</NODE>
|
|
91
|
|
|
92
|
<NODE name="similarity2actions" type="DedupSimilarityToActions">
|
|
93
|
<DESCRIPTION>export the similarity rels as Actions</DESCRIPTION>
|
|
94
|
<PARAMETERS>
|
|
95
|
<PARAM name="hadoopJob" value="dedupSimilarity2HdfsActionsJob"/>
|
|
96
|
<PARAM name="dedupConfigSequence" ref="dedupConfigSequence"/>
|
|
97
|
<PARAM name="entityType" env="entityType"/>
|
|
98
|
<PARAM name="jobParams">
|
|
99
|
<MAP>
|
|
100
|
<ENTRY key="dedup.conf" env="dedup.conf"/>
|
|
101
|
<ENTRY key="entityTypeId" env="entityTypeId"/>
|
|
102
|
<ENTRY key="entityType" env="entityType"/>
|
|
103
|
<ENTRY key="cluster" ref="cluster"/>
|
|
104
|
<ENTRY key="rawSetId" env="rawSetId"/>
|
|
105
|
<ENTRY key="mapred.output.dir" env="actionSetPath"/>
|
|
106
|
<ENTRY key="hbase.mapred.inputtable" ref="hbaseTable"/>
|
|
107
|
<ENTRY key="hbase.mapreduce.inputtable" ref="hbaseTable"/>
|
|
108
|
</MAP>
|
|
109
|
</PARAM>
|
|
110
|
</PARAMETERS>
|
|
111
|
<ARCS>
|
|
112
|
<ARC to="updateActionSets"/>
|
|
113
|
</ARCS>
|
|
114
|
</NODE>
|
|
115
|
|
|
116
|
<NODE name="updateActionSets" type="UpdateActionSets">
|
|
117
|
<DESCRIPTION>update action sets</DESCRIPTION>
|
|
118
|
<PARAMETERS/>
|
|
119
|
<ARCS>
|
|
120
|
<ARC to="doneActions"/>
|
|
121
|
</ARCS>
|
|
122
|
</NODE>
|
|
123
|
|
|
124
|
<NODE name="doneActions">
|
|
125
|
<DESCRIPTION>done actions</DESCRIPTION>
|
|
126
|
<PARAMETERS/>
|
|
127
|
<ARCS>
|
|
128
|
<ARC to="success"/>
|
|
129
|
</ARCS>
|
|
130
|
</NODE>
|
|
131
|
|
|
132
|
</WORKFLOW>
|
|
133
|
</CONFIGURATION>
|
|
134
|
</BODY>
|
|
135
|
</RESOURCE_PROFILE>
|