1
|
<RESOURCE_PROFILE>
|
2
|
<HEADER>
|
3
|
<RESOURCE_IDENTIFIER value="5e3e809c-c967-4c68-95f4-9cb8a503bbc3_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ=="/>
|
4
|
<RESOURCE_TYPE value="WorkflowTemplateDSResourceType"/>
|
5
|
<RESOURCE_KIND value="WorkflowTemplateDSResources"/>
|
6
|
<RESOURCE_URI value=""/>
|
7
|
<DATE_OF_CREATION value="2006-05-04T18:13:51.0Z"/>
|
8
|
</HEADER>
|
9
|
<BODY>
|
10
|
<CONFIGURATION>
|
11
|
|
12
|
<PARAMETERS>
|
13
|
<PARAM name="hbaseTable" description="HBase table name" required="true" type="string"/>
|
14
|
<PARAM name="cluster" description="Hadoop cluster name" required="true" type="string"/>
|
15
|
|
16
|
<PARAM name="dedupConfig" description="deduplication configuration name " required="true" type="string"/>
|
17
|
</PARAMETERS>
|
18
|
|
19
|
<WORKFLOW>
|
20
|
|
21
|
<NODE isStart="true" name="fetchRelClasses" type="FetchRelClasses">
|
22
|
<DESCRIPTION/>
|
23
|
<PARAMETERS>
|
24
|
<PARAM name="xquery" property="dnet.openaire.model.relclasses.xquery"/>
|
25
|
<PARAM name="relClassesName" value="relClasses"/>
|
26
|
</PARAMETERS>
|
27
|
<ARCS>
|
28
|
<ARC to="waitConf"/>
|
29
|
</ARCS>
|
30
|
</NODE>
|
31
|
|
32
|
|
33
|
<NODE isStart="true" name="setDedupConfigs" type="SetDedupConfiguration">
|
34
|
<DESCRIPTION>Set Dedup conf</DESCRIPTION>
|
35
|
<PARAMETERS>
|
36
|
<PARAM name="dedupConfigSequence" ref="dedupConfig"/>
|
37
|
<PARAM name="dedupConfigSequenceParam" value="dedup.conf.queue"/>
|
38
|
<PARAM name="entityTypeNameParam" value="dedup.entity.name"/>
|
39
|
<PARAM name="entityTypeIdParam" value="dedup.entity.id"/>
|
40
|
</PARAMETERS>
|
41
|
<ARCS>
|
42
|
<ARC to="waitConf"/>
|
43
|
</ARCS>
|
44
|
</NODE>
|
45
|
|
46
|
<NODE isJoin="true" name="waitConf">
|
47
|
<DESCRIPTION/>
|
48
|
<PARAMETERS/>
|
49
|
<ARCS>
|
50
|
<ARC to="dedupPhase"/>
|
51
|
</ARCS>
|
52
|
</NODE>
|
53
|
|
54
|
<NODE name="dedupPhase" type="DedupConfigurationAwareJob">
|
55
|
<DESCRIPTION>Dedup M/R job</DESCRIPTION>
|
56
|
<PARAMETERS>
|
57
|
<PARAM name="cluster" ref="cluster"/>
|
58
|
<PARAM name="hadoopJob" value="dedupPersonJob"/>
|
59
|
<PARAM name="dedupConfigurationOrchestration" env="dedup.conf.queue"/>
|
60
|
<PARAM name="jobParams">
|
61
|
<MAP>
|
62
|
<ENTRY key="hbase.mapred.inputtable" ref="hbaseTable"/>
|
63
|
<ENTRY key="hbase.mapred.outputtable" ref="hbaseTable"/>
|
64
|
<ENTRY key="hbase.mapreduce.inputtable" ref="hbaseTable"/>
|
65
|
<ENTRY key="hbase.mapreduce.outputtable" ref="hbaseTable"/>
|
66
|
<ENTRY key="entityType" env="dedup.entity.name"/>
|
67
|
<ENTRY key="entityTypeId" env="dedup.entity.id"/>
|
68
|
</MAP>
|
69
|
</PARAM>
|
70
|
</PARAMETERS>
|
71
|
<ARCS>
|
72
|
<ARC to="buildRoots"/>
|
73
|
</ARCS>
|
74
|
</NODE>
|
75
|
|
76
|
<NODE name="buildRoots" type="DedupConfigurationLoaderJob">
|
77
|
<DESCRIPTION>redirect rels</DESCRIPTION>
|
78
|
<PARAMETERS>
|
79
|
<PARAM name="hadoopJob" value="dedupBuildRootsJob"/>
|
80
|
<PARAM name="dedupConfigSequence" ref="dedupConfig"/>
|
81
|
<PARAM name="cluster" ref="cluster"/>
|
82
|
<PARAM name="jobParams">
|
83
|
<MAP>
|
84
|
<ENTRY key="hbase.mapred.inputtable" ref="hbaseTable"/>
|
85
|
<ENTRY key="hbase.mapred.outputtable" ref="hbaseTable"/>
|
86
|
<ENTRY key="hbase.mapreduce.inputtable" ref="hbaseTable"/>
|
87
|
<ENTRY key="hbase.mapreduce.outputtable" ref="hbaseTable"/>
|
88
|
<ENTRY key="entityType" env="dedup.entity.name"/>
|
89
|
<ENTRY key="entityTypeId" env="dedup.entity.id"/>
|
90
|
<ENTRY key="relClasses" env="relClasses"/>
|
91
|
</MAP>
|
92
|
</PARAM>
|
93
|
</PARAMETERS>
|
94
|
<ARCS>
|
95
|
<ARC to="success"/>
|
96
|
</ARCS>
|
97
|
</NODE>
|
98
|
|
99
|
</WORKFLOW>
|
100
|
</CONFIGURATION>
|
101
|
</BODY>
|
102
|
</RESOURCE_PROFILE>
|