1 |
37369
|
claudio.at
|
<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="tableParam" managedBy="system">tableName</PARAM>
|
29 |
|
|
</PARAMETERS>
|
30 |
|
|
<ARCS>
|
31 |
|
|
<ARC to="waitConf" />
|
32 |
|
|
</ARCS>
|
33 |
|
|
</NODE>
|
34 |
|
|
|
35 |
|
|
<NODE name="setDedupConfigs" type="SetDedupConfiguration" isStart="true">
|
36 |
|
|
<DESCRIPTION>Set Dedup conf</DESCRIPTION>
|
37 |
|
|
<PARAMETERS>
|
38 |
|
|
<PARAM required="true" type="string" name="dedupConfigSequence" managedBy="user">dedup-similarity-person</PARAM>
|
39 |
|
|
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM>
|
40 |
|
|
</PARAMETERS>
|
41 |
|
|
<ARCS>
|
42 |
|
|
<ARC to="waitConf" />
|
43 |
|
|
</ARCS>
|
44 |
|
|
</NODE>
|
45 |
|
|
|
46 |
|
|
<NODE name="waitConf" isJoin="true">
|
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 managedBy="system" name="hadoopJob" required="true" type="string">dedupPersonJob</PARAM>
|
58 |
|
|
<PARAM managedBy="system" name="dedupConfigSequenceParam" type="string" required="true">dedup.conf.queue</PARAM>
|
59 |
|
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
60 |
|
|
{
|
61 |
|
|
'entityTypeId' : 'entityTypeId',
|
62 |
|
|
'entityType' : 'entityType',
|
63 |
|
|
'cluster' : 'cluster',
|
64 |
|
|
'hbase.mapred.inputtable' : 'tableName',
|
65 |
|
|
'hbase.mapred.outputtable' : 'tableName',
|
66 |
|
|
'hbase.mapreduce.inputtable' : 'tableName',
|
67 |
|
|
'hbase.mapreduce.outputtable' : 'tableName'
|
68 |
|
|
}
|
69 |
|
|
</PARAM>
|
70 |
|
|
</PARAMETERS>
|
71 |
|
|
<ARCS>
|
72 |
|
|
<ARC to="buildRoots"/>
|
73 |
|
|
</ARCS>
|
74 |
|
|
</NODE>
|
75 |
|
|
<NODE name="buildRoots" type="SubmitHadoopJob">
|
76 |
|
|
<DESCRIPTION>redirect rels</DESCRIPTION>
|
77 |
|
|
<PARAMETERS>
|
78 |
|
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupBuildRootsJob</PARAM>
|
79 |
|
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
80 |
|
|
{
|
81 |
|
|
'dedup.conf' : 'dedup.conf',
|
82 |
|
|
'relClasses' : 'relClasses',
|
83 |
|
|
'entityTypeId' : 'entityTypeId',
|
84 |
|
|
'entityType' : 'entityType',
|
85 |
|
|
'cluster' : 'cluster',
|
86 |
|
|
'hbase.mapred.inputtable' : 'tableName',
|
87 |
|
|
'hbase.mapreduce.inputtable' : 'tableName',
|
88 |
|
|
'hbase.mapred.outputtable' : 'tableName',
|
89 |
|
|
'hbase.mapreduce.outputtable' : 'tableName'
|
90 |
|
|
}
|
91 |
|
|
</PARAM>
|
92 |
|
|
</PARAMETERS>
|
93 |
|
|
<ARCS>
|
94 |
|
|
<ARC to="success"/>
|
95 |
|
|
</ARCS>
|
96 |
|
|
</NODE>
|
97 |
|
|
</CONFIGURATION>
|
98 |
|
|
<STATUS>
|
99 |
|
|
<LAST_EXECUTION_ID>wf_20150513_160956_878</LAST_EXECUTION_ID>
|
100 |
|
|
<LAST_EXECUTION_DATE>2015-05-13T16:58:47+00:00</LAST_EXECUTION_DATE>
|
101 |
|
|
<LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
|
102 |
|
|
<LAST_EXECUTION_ERROR/>
|
103 |
|
|
</STATUS>
|
104 |
|
|
</BODY>
|
105 |
|
|
</RESOURCE_PROFILE>
|