1
|
<RESOURCE_PROFILE>
|
2
|
<HEADER>
|
3
|
<RESOURCE_IDENTIFIER value="f8d1f8f1-2cf4-4f84-af70-ceb11ec120de_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
|
4
|
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
|
5
|
<RESOURCE_KIND value="WorkflowDSResources"/>
|
6
|
<RESOURCE_URI value=""/>
|
7
|
<DATE_OF_CREATION value="2015-11-12T17:19:38+00:00"/>
|
8
|
</HEADER>
|
9
|
<BODY>
|
10
|
<WORKFLOW_NAME>HBase to HDFS</WORKFLOW_NAME>
|
11
|
<WORKFLOW_TYPE>Data Provision</WORKFLOW_TYPE>
|
12
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
13
|
<CONFIGURATION start="manual">
|
14
|
|
15
|
<!-- PREPARE NODES -->
|
16
|
<NODE isStart="true" name="setInfo" type="SetFormatInfo">
|
17
|
<DESCRIPTION>set mdformat, layout, interpretation</DESCRIPTION>
|
18
|
<PARAMETERS>
|
19
|
<PARAM managedBy="user" name="format" required="true" type="string">TMF</PARAM>
|
20
|
<PARAM managedBy="user" name="layout" required="true" type="string">index</PARAM>
|
21
|
<PARAM managedBy="user" name="interpretation" required="true" type="string">openaire</PARAM>
|
22
|
</PARAMETERS>
|
23
|
<ARCS>
|
24
|
<ARC to="findIndex"/>
|
25
|
</ARCS>
|
26
|
</NODE>
|
27
|
<NODE isStart="true" name="fetchRelClasses" type="FetchRelClasses">
|
28
|
<DESCRIPTION/>
|
29
|
<PARAMETERS>
|
30
|
<PARAM managedBy="system" name="relClassesProperty" required="true" type="string">dnet.openaire.model.relclasses.xquery</PARAM>
|
31
|
<PARAM managedBy="system" name="relClassesName" required="true" type="string">relClasses</PARAM>
|
32
|
</PARAMETERS>
|
33
|
<ARCS>
|
34
|
<ARC to="groupEntities"/>
|
35
|
</ARCS>
|
36
|
</NODE>
|
37
|
<NODE isStart="true" name="fetchContexts" type="LoadContextsJob">
|
38
|
<DESCRIPTION/>
|
39
|
<PARAMETERS/>
|
40
|
<ARCS>
|
41
|
<ARC to="groupEntities"/>
|
42
|
</ARCS>
|
43
|
</NODE>
|
44
|
<NODE isStart="true" name="fetchEntityLinks" type="LoadEntityLinksJob">
|
45
|
<DESCRIPTION/>
|
46
|
<PARAMETERS/>
|
47
|
<ARCS>
|
48
|
<ARC to="groupEntities"/>
|
49
|
</ARCS>
|
50
|
</NODE>
|
51
|
|
52
|
<!-- UPDATE INDEX -->
|
53
|
<NODE name="findIndex" type="FindIndex">
|
54
|
<DESCRIPTION/>
|
55
|
<PARAMETERS/>
|
56
|
<ARCS>
|
57
|
<ARC name="found" to="prepareIndexing"/>
|
58
|
<ARC name="notFound" to="createIndex"/>
|
59
|
</ARCS>
|
60
|
</NODE>
|
61
|
<NODE name="createIndex" type="CreateIndex">
|
62
|
<DESCRIPTION/>
|
63
|
<PARAMETERS/>
|
64
|
<ARCS>
|
65
|
<ARC to="prepareIndexing"/>
|
66
|
</ARCS>
|
67
|
</NODE>
|
68
|
<NODE name="prepareIndexing" type="PrepareIndexJob">
|
69
|
<DESCRIPTION>Prepare indexing</DESCRIPTION>
|
70
|
<PARAMETERS>
|
71
|
<PARAM managedBy="system" name="outputRecordsPathParam" required="true" type="string">hdfsRecordsPath</PARAM>
|
72
|
<PARAM managedBy="system" name="rottenRecordsPathParam" required="true" type="string">rottenRecordsPath</PARAM>
|
73
|
<PARAM managedBy="system" name="layoutToRecordStylesheet" required="true"
|
74
|
type="string">/eu/dnetlib/msro/openaireplus/workflows/index/openaireLayoutToRecordStylesheet.xsl
|
75
|
</PARAM>
|
76
|
<PARAM managedBy="system" name="oafSchemaLocationProperty" required="true" type="string">oaf.schema.location</PARAM>
|
77
|
</PARAMETERS>
|
78
|
<ARCS>
|
79
|
<ARC to="cleanupXml"/>
|
80
|
<ARC to="cleanupRotten"/>
|
81
|
</ARCS>
|
82
|
</NODE>
|
83
|
<NODE name="cleanupXml" type="DeleteHdfsPathJob">
|
84
|
<DESCRIPTION>hdfs cleanup (xml)</DESCRIPTION>
|
85
|
<PARAMETERS>
|
86
|
<PARAM managedBy="system" name="cluster" required="true" type="string">DM</PARAM>
|
87
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
88
|
{
|
89
|
'path' : 'hdfsRecordsPath'
|
90
|
}
|
91
|
</PARAM>
|
92
|
</PARAMETERS>
|
93
|
<ARCS>
|
94
|
<ARC to="groupEntities"/>
|
95
|
</ARCS>
|
96
|
</NODE>
|
97
|
<NODE name="cleanupRotten" type="DeleteHdfsPathJob">
|
98
|
<DESCRIPTION>hdfs cleanup (rotten)</DESCRIPTION>
|
99
|
<PARAMETERS>
|
100
|
<PARAM managedBy="system" name="cluster" required="true" type="string">DM</PARAM>
|
101
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
102
|
{
|
103
|
'path' : 'rottenRecordsPath'
|
104
|
}
|
105
|
</PARAM>
|
106
|
</PARAMETERS>
|
107
|
<ARCS>
|
108
|
<ARC to="groupEntities"/>
|
109
|
</ARCS>
|
110
|
</NODE>
|
111
|
<NODE isJoin="true" name="groupEntities" type="SubmitHadoopJob">
|
112
|
<DESCRIPTION>M/R group entities</DESCRIPTION>
|
113
|
<PARAMETERS>
|
114
|
<PARAM managedBy="system" name="cluster" required="true" type="string">DM</PARAM>
|
115
|
<PARAM managedBy="system" name="hadoopJob" required="true" type="string">prepareIndexDataJob</PARAM>
|
116
|
<PARAM managedBy="system" name="sysParams" required="true" type="string">
|
117
|
{
|
118
|
'hbase.mapred.inputtable' : 'hbase.mapred.datatable',
|
119
|
'hbase.mapreduce.inputtable' : 'hbase.mapred.datatable'
|
120
|
}
|
121
|
</PARAM>
|
122
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
123
|
{
|
124
|
'mapred.output.dir' : 'hdfsRecordsPath',
|
125
|
'index.entity.links' : 'index.entity.links',
|
126
|
'oaf.schema.location' : 'oaf.schema.location',
|
127
|
'contextmap' : 'contextmap',
|
128
|
'relClasses' : 'relClasses'
|
129
|
}
|
130
|
</PARAM>
|
131
|
</PARAMETERS>
|
132
|
<ARCS>
|
133
|
<ARC to="success"/>
|
134
|
</ARCS>
|
135
|
</NODE>
|
136
|
</CONFIGURATION>
|
137
|
<STATUS>
|
138
|
</STATUS>
|
139
|
</BODY>
|
140
|
</RESOURCE_PROFILE>
|