|
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="35044412-9bee-479b-a8d5-e15ba2fcad3a_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>Index from HBASE</WORKFLOW_NAME>
|
|
12
|
<WORKFLOW_TYPE>Provision</WORKFLOW_TYPE>
|
|
13
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
|
14
|
<CONFIGURATION start="auto">
|
|
15
|
<NODE name="findIndex" type="FindIndex" isStart="true">
|
|
16
|
<DESCRIPTION />
|
|
17
|
<PARAMETERS>
|
|
18
|
<PARAM name="mdFormat" type="string" managedBy="user" required="true">DMF</PARAM>
|
|
19
|
<PARAM name="layout" type="string" managedBy="system" required="true">index</PARAM>
|
|
20
|
<PARAM name="interpretation" type="string" managedBy="system" required="true">openaire</PARAM>
|
|
21
|
</PARAMETERS>
|
|
22
|
<ARCS>
|
|
23
|
<ARC name="found" to="prepare" />
|
|
24
|
<ARC name="notFound" to="createIndex" />
|
|
25
|
</ARCS>
|
|
26
|
</NODE>
|
|
27
|
<NODE name="createIndex" type="CreateIndex">
|
|
28
|
<DESCRIPTION />
|
|
29
|
<PARAMETERS />
|
|
30
|
<ARCS>
|
|
31
|
<ARC to="prepare" />
|
|
32
|
</ARCS>
|
|
33
|
</NODE>
|
|
34
|
<NODE name="prepare" type="PrepareIndexJob">
|
|
35
|
<DESCRIPTION>Prepare indexing</DESCRIPTION>
|
|
36
|
<PARAMETERS>
|
|
37
|
<PARAM name="outputRecordsPathParam" type="string" managedBy="system" required="true">outputRecordsPath</PARAM>
|
|
38
|
<PARAM name="rottenRecordsPathParam" type="string" managedBy="system" required="true">rottenRecordsPath</PARAM>
|
|
39
|
<PARAM required="true" type="string" name="layoutToRecordStylesheet" managedBy="system">/eu/dnetlib/msro/openaireplus/workflows/index/openaireLayoutToRecordStylesheet.xsl</PARAM>
|
|
40
|
</PARAMETERS>
|
|
41
|
<ARCS>
|
|
42
|
<ARC to="cleanupXml" />
|
|
43
|
<ARC to="cleanupRotten" />
|
|
44
|
</ARCS>
|
|
45
|
</NODE>
|
|
46
|
<NODE name="cleanupXml" type="DeleteHdfsPathJob">
|
|
47
|
<DESCRIPTION>hdfs cleanup (xml)</DESCRIPTION>
|
|
48
|
<PARAMETERS>
|
|
49
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
|
50
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
|
51
|
{
|
|
52
|
'path' : 'outputRecordsPath'
|
|
53
|
}
|
|
54
|
</PARAM>
|
|
55
|
</PARAMETERS>
|
|
56
|
<ARCS>
|
|
57
|
<ARC to="group" />
|
|
58
|
</ARCS>
|
|
59
|
</NODE>
|
|
60
|
<NODE name="cleanupRotten" type="DeleteHdfsPathJob">
|
|
61
|
<DESCRIPTION>hdfs cleanup (rotten)</DESCRIPTION>
|
|
62
|
<PARAMETERS>
|
|
63
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
|
64
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
|
65
|
{
|
|
66
|
'path' : 'rottenRecordsPath'
|
|
67
|
}
|
|
68
|
</PARAM>
|
|
69
|
</PARAMETERS>
|
|
70
|
<ARCS>
|
|
71
|
<ARC to="group" />
|
|
72
|
<ARC to="updateStats" />
|
|
73
|
</ARCS>
|
|
74
|
</NODE>
|
|
75
|
<NODE name="group" type="SubmitHadoopJob" isJoin="true">
|
|
76
|
<DESCRIPTION>M/R group entities</DESCRIPTION>
|
|
77
|
<PARAMETERS>
|
|
78
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
|
79
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">prepareIndexDataJob</PARAM>
|
|
80
|
<PARAM required="true" type="string" name="sysParams" managedBy="system">
|
|
81
|
{
|
|
82
|
'hbase.mapred.inputtable' : 'hbase.mapred.datatable',
|
|
83
|
'hbase.mapreduce.inputtable' : 'hbase.mapred.datatable'
|
|
84
|
}
|
|
85
|
</PARAM>
|
|
86
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
|
87
|
{
|
|
88
|
'mapred.output.dir' : 'outputRecordsPath',
|
|
89
|
'index.entity.links' : 'index.entity.links',
|
|
90
|
'contextmap' : 'contextmap'
|
|
91
|
}
|
|
92
|
</PARAM>
|
|
93
|
</PARAMETERS>
|
|
94
|
<ARCS>
|
|
95
|
<ARC to="updateIndex" />
|
|
96
|
</ARCS>
|
|
97
|
</NODE>
|
|
98
|
<NODE name="updateIndex" type="SubmitHadoopJob">
|
|
99
|
<DESCRIPTION>M/O index records</DESCRIPTION>
|
|
100
|
<PARAMETERS>
|
|
101
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
|
102
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">indexFeedJob</PARAM>
|
|
103
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
|
104
|
{
|
|
105
|
'mapred.input.dir' : 'outputRecordsPath',
|
|
106
|
'mapred.output.dir' : 'rottenRecordsPath',
|
|
107
|
'index.xslt' : 'index.xslt',
|
|
108
|
'index.solr.url.zk' : 'index.solr.url.zk',
|
|
109
|
'index.solr.url.list' : 'index.solr.url.list',
|
|
110
|
'index.solr.url.local' : 'index.solr.url.local',
|
|
111
|
'index.solr.collection' : 'index.solr.collection',
|
|
112
|
'index.buffer.flush.threshold' : 'index.buffer.flush.threshold',
|
|
113
|
'index.shutdown.wait.time' : 'index.shutdown.wait.time',
|
|
114
|
'index.solr.local.feeding' : 'index.solr.local.feeding',
|
|
115
|
'index.solr.sim.mode' : 'index.solr.sim.mode',
|
|
116
|
'index.feed.timestamp' : 'index.feed.timestamp'
|
|
117
|
}
|
|
118
|
</PARAM>
|
|
119
|
</PARAMETERS>
|
|
120
|
<ARCS>
|
|
121
|
<ARC to="sync" />
|
|
122
|
</ARCS>
|
|
123
|
</NODE>
|
|
124
|
<NODE name="updateStats" type="SubmitHadoopMapReduce">
|
|
125
|
<DESCRIPTION>update stats</DESCRIPTION>
|
|
126
|
<PARAMETERS />
|
|
127
|
<ARCS>
|
|
128
|
<ARC to="sync" />
|
|
129
|
</ARCS>
|
|
130
|
</NODE>
|
|
131
|
<NODE name="sync" isJoin="true">
|
|
132
|
<DESCRIPTION />
|
|
133
|
<PARAMETERS />
|
|
134
|
<ARCS>
|
|
135
|
<ARC to="finalize" />
|
|
136
|
</ARCS>
|
|
137
|
</NODE>
|
|
138
|
<NODE name="finalize" type="FinalizeIndexFeeding">
|
|
139
|
<DESCRIPTION>commit changes</DESCRIPTION>
|
|
140
|
<PARAMETERS />
|
|
141
|
<ARCS>
|
|
142
|
<ARC to="updateDs" />
|
|
143
|
</ARCS>
|
|
144
|
</NODE>
|
|
145
|
<NODE name="updateDs" type="IndexDsUpdateJob">
|
|
146
|
<DESCRIPTION>update DS</DESCRIPTION>
|
|
147
|
<PARAMETERS />
|
|
148
|
<ARCS>
|
|
149
|
<ARC to="switch" />
|
|
150
|
</ARCS>
|
|
151
|
</NODE>
|
|
152
|
<NODE name="switch" type="SwitchIndex">
|
|
153
|
<DESCRIPTION>switch index</DESCRIPTION>
|
|
154
|
<PARAMETERS />
|
|
155
|
<ARCS>
|
|
156
|
<ARC to="success" />
|
|
157
|
</ARCS>
|
|
158
|
</NODE>
|
|
159
|
</CONFIGURATION>
|
|
160
|
<STATUS />
|
|
161
|
</BODY>
|
|
162
|
</RESOURCE_PROFILE>
|
|
163
|
|
|
164
|
|