Project

General

Profile

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="cc6cf74c-5fef-4cdf-9f93-bd02b70709d2_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 persons</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">person</PARAM>
29
				</PARAMETERS>
30
				<ARCS>
31
					<ARC to="dedupPhase" />
32
				</ARCS>
33
			</NODE>
34
			
35
			<NODE name="dedupPhase" type="SubmitHadoopJob" isJoin="true">
36
				<DESCRIPTION>Dedup M/R job</DESCRIPTION>
37
				<PARAMETERS>
38
					<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupPersonJob</PARAM>
39
					<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
40
					<PARAM required="true" type="string" name="sysParams" managedBy="system">
41
						{ 	
42
							'hbase.mapred.inputtable' : 'hbase.mapred.datatable', 
43
							'hbase.mapred.outputtable' : 'hbase.mapred.datatable', 
44
							'hbase.mapreduce.inputtable' : 'hbase.mapred.datatable', 
45
							'hbase.mapreduce.outputtable' : 'hbase.mapred.datatable'
46
						}
47
					</PARAM>
48
					<PARAM required="true" type="string" name="envParams" managedBy="system">
49
						{ 	
50
							'dedup.pace.conf' : 'dedup.pace.conf',
51
							'dedup.wf.conf' : 'dedup.wf.conf'
52
						}
53
					</PARAM>					
54
					<PARAM required="true" type="string" name="params" managedBy="system">
55
						{ 	
56
							'entityTypeId' : '30',
57
							'entityType' : 'person'							 
58
						}
59
					</PARAM>
60
				</PARAMETERS>
61
				<ARCS>
62
					<ARC to="buildRoots" />
63
				</ARCS>
64
			</NODE>
65
			<NODE name="buildRoots" type="SubmitHadoopJob">
66
				<DESCRIPTION>redirect rels</DESCRIPTION>
67
				<PARAMETERS>
68
					<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupBuildRootsJob</PARAM>
69
					<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
70
					<PARAM required="true" type="string" name="sysParams" managedBy="system">
71
						{ 	
72
							'hbase.mapred.inputtable' : 'hbase.mapred.datatable', 
73
							'hbase.mapred.outputtable' : 'hbase.mapred.datatable', 
74
							'hbase.mapreduce.inputtable' : 'hbase.mapred.datatable', 
75
							'hbase.mapreduce.outputtable' : 'hbase.mapred.datatable'
76
						}
77
					</PARAM>
78
					<PARAM required="true" type="string" name="envParams" managedBy="system">
79
						{ 	
80
							'dedup.pace.conf' : 'dedup.pace.conf',
81
							'dedup.wf.conf' : 'dedup.wf.conf',
82
							'relClasses' : 'relClasses'
83
						}
84
					</PARAM>					
85
					<PARAM required="true" type="string" name="params" managedBy="system">
86
						{ 	
87
							'entityTypeId' : '30'
88
						}
89
					</PARAM>				
90
				</PARAMETERS>
91
				<ARCS>
92
					<ARC to="success" />
93
				</ARCS>
94
			</NODE>
95
		</CONFIGURATION>
96
		<STATUS />
97
	</BODY>
98
</RESOURCE_PROFILE>
(2-2/5)