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="07093922-e99e-41fc-b931-41019a82f736_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>Reindex Deduplication</WORKFLOW_NAME>
|
12
|
<WORKFLOW_TYPE>Deduplication</WORKFLOW_TYPE>
|
13
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
14
|
<CONFIGURATION start="manual">
|
15
|
|
16
|
|
17
|
<NODE name="setDedupConfigs" type="SetDedupConfiguration" isStart="true">
|
18
|
<DESCRIPTION>Set Dedup conf</DESCRIPTION>
|
19
|
<PARAMETERS>
|
20
|
<PARAM function="obtainValues('dedupOrchestrations', {})" required="true" type="string" name="dedupConfigSequence" managedBy="user"></PARAM>
|
21
|
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM>
|
22
|
</PARAMETERS>
|
23
|
<ARCS>
|
24
|
<ARC to="checkConf" />
|
25
|
</ARCS>
|
26
|
</NODE>
|
27
|
<NODE name="hadoopConfig" type="SetClusterAndTable" isStart="true">
|
28
|
<DESCRIPTION>Set table name</DESCRIPTION>
|
29
|
<PARAMETERS>
|
30
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
31
|
<PARAM required="true" type="string" name="tableParam" managedBy="system">tableName</PARAM>
|
32
|
</PARAMETERS>
|
33
|
<ARCS>
|
34
|
<ARC to="checkConf" />
|
35
|
</ARCS>
|
36
|
</NODE>
|
37
|
<NODE name="checkConf" type="DedupCheckConfiguration" isJoin="true">
|
38
|
<DESCRIPTION/>
|
39
|
<PARAMETERS>
|
40
|
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM>
|
41
|
</PARAMETERS>
|
42
|
<ARCS>
|
43
|
<ARC to="prepareActionSet" />
|
44
|
</ARCS>
|
45
|
</NODE>
|
46
|
|
47
|
<NODE name="prepareActionSet" type="PrepareConfiguredActionSet">
|
48
|
<DESCRIPTION>prepare action sets</DESCRIPTION>
|
49
|
<PARAMETERS>
|
50
|
<PARAM required="true" type="string" name="dedupConfigSequenceParam" managedBy="system">dedup.conf.queue</PARAM>
|
51
|
<PARAM required="true" type="string" name="jobProperty" managedBy="system">rawSetId</PARAM>
|
52
|
</PARAMETERS>
|
53
|
<ARCS>
|
54
|
<ARC to="findIndex" />
|
55
|
</ARCS>
|
56
|
</NODE>
|
57
|
|
58
|
<NODE name="findIndex" type="FindIndex">
|
59
|
<DESCRIPTION />
|
60
|
<PARAMETERS>
|
61
|
<PARAM name="mdFormat" type="string" managedBy="system" required="true">OPENAIRE</PARAM>
|
62
|
<PARAM name="layout" type="string" managedBy="system" required="true">index</PARAM>
|
63
|
<PARAM name="interpretation" type="string" managedBy="system" required="true">dedup</PARAM>
|
64
|
</PARAMETERS>
|
65
|
<ARCS>
|
66
|
<ARC name="found" to="prepareIndex" />
|
67
|
<ARC name="notFound" to="createIndex" />
|
68
|
</ARCS>
|
69
|
</NODE>
|
70
|
|
71
|
<NODE name="createIndex" type="CreateIndex">
|
72
|
<DESCRIPTION />
|
73
|
<PARAMETERS />
|
74
|
<ARCS>
|
75
|
<ARC to="prepareIndex" />
|
76
|
</ARCS>
|
77
|
</NODE>
|
78
|
|
79
|
<NODE name="prepareIndex" type="PrepareDedupIndexJob">
|
80
|
<DESCRIPTION>Prepare indexing</DESCRIPTION>
|
81
|
<PARAMETERS>
|
82
|
<PARAM name="rottenRecordsPathParam" type="string" required="true" managedBy="system">rottenRecordsPath</PARAM>
|
83
|
<PARAM required="true" type="string" name="dedupConfig" managedBy="system">
|
84
|
{
|
85
|
'dedupConfig' : 'dedup.conf'
|
86
|
}
|
87
|
</PARAM>
|
88
|
</PARAMETERS>
|
89
|
<ARCS>
|
90
|
<ARC to="cleanupRotten" />
|
91
|
</ARCS>
|
92
|
</NODE>
|
93
|
|
94
|
<NODE name="cleanupRotten" type="DeleteHdfsPathJob">
|
95
|
<DESCRIPTION>hdfs cleanup (rotten)</DESCRIPTION>
|
96
|
<PARAMETERS>
|
97
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
98
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
99
|
{
|
100
|
'path' : 'rottenRecordsPath'
|
101
|
}
|
102
|
</PARAM>
|
103
|
</PARAMETERS>
|
104
|
<ARCS>
|
105
|
<ARC to="updateIndex" />
|
106
|
</ARCS>
|
107
|
</NODE>
|
108
|
|
109
|
<NODE name="updateIndex" type="SubmitHadoopJob">
|
110
|
<DESCRIPTION>M/O index records</DESCRIPTION>
|
111
|
<PARAMETERS>
|
112
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">dedupIndexFeedJob</PARAM>
|
113
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
114
|
{
|
115
|
'mapred.output.dir' : 'rottenRecordsPath',
|
116
|
'index.fields' : 'index.fields',
|
117
|
'index.solr.url' : 'index.solr.url',
|
118
|
'index.solr.collection' : 'index.solr.collection',
|
119
|
'index.buffer.flush.threshold' : 'index.buffer.flush.threshold',
|
120
|
'index.shutdown.wait.time' : 'index.shutdown.wait.time',
|
121
|
'index.solr.sim.mode' : 'index.solr.sim.mode',
|
122
|
'index.feed.timestamp' : 'index.feed.timestamp',
|
123
|
'entityTypeId' : 'entityTypeId',
|
124
|
'entityType' : 'entityType',
|
125
|
'actionset' : 'actionset',
|
126
|
'cluster' : 'cluster'
|
127
|
}
|
128
|
</PARAM>
|
129
|
<PARAM required="true" type="string" name="sysParams" managedBy="system">
|
130
|
{
|
131
|
'hbase.mapred.inputtable' : 'hbase.mapred.datatable',
|
132
|
'hbase.mapreduce.inputtable' : 'hbase.mapred.datatable'
|
133
|
}
|
134
|
</PARAM>
|
135
|
</PARAMETERS>
|
136
|
<ARCS>
|
137
|
<ARC to="finalize" />
|
138
|
</ARCS>
|
139
|
</NODE>
|
140
|
|
141
|
<NODE name="finalize" type="FinalizeDedupIndexFeeding">
|
142
|
<DESCRIPTION>commit changes</DESCRIPTION>
|
143
|
<PARAMETERS />
|
144
|
<ARCS>
|
145
|
<ARC to="updateDs" />
|
146
|
</ARCS>
|
147
|
</NODE>
|
148
|
|
149
|
<NODE name="updateDs" type="IndexDsUpdateJob">
|
150
|
<DESCRIPTION>update DS</DESCRIPTION>
|
151
|
<PARAMETERS />
|
152
|
<ARCS>
|
153
|
<ARC to="success" />
|
154
|
</ARCS>
|
155
|
</NODE>
|
156
|
|
157
|
|
158
|
</CONFIGURATION>
|
159
|
<STATUS />
|
160
|
</BODY>
|
161
|
</RESOURCE_PROFILE>
|