Project

General

Profile

1
<RESOURCE_PROFILE>
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="2015-08-29T14:48:32+00:00"/>
8
	</HEADER>
9
	<BODY>
10
		<WORKFLOW_NAME>Dedup persons (Online)</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 isStart="true" name="hadoopConfig" type="SetClusterAndTable">
25
				<DESCRIPTION>Set table name</DESCRIPTION>
26
				<PARAMETERS>
27
					<PARAM managedBy="system" name="cluster" required="true" type="string">DM</PARAM>
28
					<PARAM managedBy="system" name="tableParam" required="true" type="string">tableName</PARAM>
29
				</PARAMETERS>
30
				<ARCS>
31
					<ARC to="waitConf"/>
32
				</ARCS>
33
			</NODE>
34
			<NODE isStart="true" name="setDedupConfigs" type="SetDedupConfiguration">
35
				<DESCRIPTION>Set Dedup conf</DESCRIPTION>
36
				<PARAMETERS>
37
					<PARAM managedBy="user" name="dedupConfigSequence" required="true" type="string">dedup-similarity-person</PARAM>
38
					<PARAM managedBy="system" name="dedupConfigSequenceParam" required="true" type="string">dedup.conf.queue</PARAM>
39
				</PARAMETERS>
40
				<ARCS>
41
					<ARC to="waitConf"/>
42
				</ARCS>
43
			</NODE>
44
			<NODE isJoin="true" name="waitConf">
45
				<DESCRIPTION/>
46
				<PARAMETERS/>
47
				<ARCS>
48
					<ARC to="dedupPhase"/>
49
				</ARCS>
50
			</NODE>
51
			<NODE name="dedupPhase" type="DedupConfigurationAwareJob">
52
				<DESCRIPTION>Dedup M/R job</DESCRIPTION>
53
				<PARAMETERS>
54
					<PARAM managedBy="system" name="hadoopJob" required="true" type="string">dedupPersonJob</PARAM>
55
					<PARAM managedBy="system" name="dedupConfigSequenceParam" required="true" type="string">dedup.conf.queue</PARAM>
56
					<PARAM managedBy="system" name="envParams" required="true" type="string">
57
						{
58
						'entityTypeId' : 'entityTypeId',
59
						'entityType' : 'entityType',
60
						'cluster' : 'cluster',
61
						'hbase.mapred.inputtable' : 'tableName',
62
						'hbase.mapred.outputtable' : 'tableName',
63
						'hbase.mapreduce.inputtable' : 'tableName',
64
						'hbase.mapreduce.outputtable' : 'tableName'
65
						}
66
					</PARAM>
67
				</PARAMETERS>
68
				<ARCS>
69
					<ARC to="buildRoots"/>
70
				</ARCS>
71
			</NODE>
72
			<NODE name="buildRoots" type="SubmitHadoopJob">
73
				<DESCRIPTION>redirect rels</DESCRIPTION>
74
				<PARAMETERS>
75
					<PARAM managedBy="system" name="hadoopJob" required="true" type="string">dedupBuildRootsJob</PARAM>
76
					<PARAM managedBy="system" name="envParams" required="true" type="string">
77
						{
78
						'dedup.conf' : 'dedup.conf',
79
						'relClasses' : 'relClasses',
80
						'entityTypeId' : 'entityTypeId',
81
						'entityType' : 'entityType',
82
						'cluster' : 'cluster',
83
						'hbase.mapred.inputtable' : 'tableName',
84
						'hbase.mapreduce.inputtable' : 'tableName',
85
						'hbase.mapred.outputtable' : 'tableName',
86
						'hbase.mapreduce.outputtable' : 'tableName'
87
						}
88
					</PARAM>
89
				</PARAMETERS>
90
				<ARCS>
91
					<ARC to="success"/>
92
				</ARCS>
93
			</NODE>
94
		</CONFIGURATION>
95
		<STATUS>
96
			<LAST_EXECUTION_ID>wf_20151020_211958_695</LAST_EXECUTION_ID>
97
			<LAST_EXECUTION_DATE>2015-10-20T22:32:39+00:00</LAST_EXECUTION_DATE>
98
			<LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
99
			<LAST_EXECUTION_ERROR/>
100
		</STATUS>
101
	</BODY>
102
</RESOURCE_PROFILE>
(4-4/8)