1 |
26600
|
sandro.lab
|
<?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 |
27202
|
claudio.at
|
<WORKFLOW_TYPE>Provision</WORKFLOW_TYPE>
|
13 |
26600
|
sandro.lab
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
14 |
28824
|
claudio.at
|
<CONFIGURATION start="manual">
|
15 |
|
|
<NODE name="fetchRelClasses" type="FetchRelClasses" isStart="true">
|
16 |
|
|
<DESCRIPTION />
|
17 |
|
|
<PARAMETERS>
|
18 |
|
|
<PARAM name="relClassesProperty" type="string" managedBy="system" required="true">dnet.openaire.model.relclasses.xquery</PARAM>
|
19 |
|
|
<PARAM name="relClassesName" type="string" managedBy="system" required="true">relClasses</PARAM>
|
20 |
|
|
</PARAMETERS>
|
21 |
|
|
<ARCS>
|
22 |
|
|
<ARC to="group" />
|
23 |
|
|
</ARCS>
|
24 |
|
|
</NODE>
|
25 |
29397
|
claudio.at
|
<NODE name="fetchContexts" type="LoadContextsJob" isStart="true">
|
26 |
|
|
<DESCRIPTION />
|
27 |
|
|
<PARAMETERS/>
|
28 |
|
|
<ARCS>
|
29 |
|
|
<ARC to="group" />
|
30 |
|
|
</ARCS>
|
31 |
|
|
</NODE>
|
32 |
|
|
<NODE name="fetchEntityLinks" type="LoadEntityLinksJob" isStart="true">
|
33 |
|
|
<DESCRIPTION />
|
34 |
|
|
<PARAMETERS/>
|
35 |
|
|
<ARCS>
|
36 |
|
|
<ARC to="group" />
|
37 |
|
|
</ARCS>
|
38 |
|
|
</NODE>
|
39 |
26600
|
sandro.lab
|
<NODE name="findIndex" type="FindIndex" isStart="true">
|
40 |
|
|
<DESCRIPTION />
|
41 |
|
|
<PARAMETERS>
|
42 |
|
|
<PARAM name="mdFormat" type="string" managedBy="user" required="true">DMF</PARAM>
|
43 |
|
|
<PARAM name="layout" type="string" managedBy="system" required="true">index</PARAM>
|
44 |
|
|
<PARAM name="interpretation" type="string" managedBy="system" required="true">openaire</PARAM>
|
45 |
|
|
</PARAMETERS>
|
46 |
|
|
<ARCS>
|
47 |
|
|
<ARC name="found" to="prepare" />
|
48 |
|
|
<ARC name="notFound" to="createIndex" />
|
49 |
|
|
</ARCS>
|
50 |
|
|
</NODE>
|
51 |
|
|
<NODE name="createIndex" type="CreateIndex">
|
52 |
|
|
<DESCRIPTION />
|
53 |
|
|
<PARAMETERS />
|
54 |
|
|
<ARCS>
|
55 |
|
|
<ARC to="prepare" />
|
56 |
|
|
</ARCS>
|
57 |
|
|
</NODE>
|
58 |
|
|
<NODE name="prepare" type="PrepareIndexJob">
|
59 |
|
|
<DESCRIPTION>Prepare indexing</DESCRIPTION>
|
60 |
|
|
<PARAMETERS>
|
61 |
|
|
<PARAM name="outputRecordsPathParam" type="string" managedBy="system" required="true">outputRecordsPath</PARAM>
|
62 |
|
|
<PARAM name="rottenRecordsPathParam" type="string" managedBy="system" required="true">rottenRecordsPath</PARAM>
|
63 |
28824
|
claudio.at
|
<PARAM name="layoutToRecordStylesheet" type="string" managedBy="system" required="true">/eu/dnetlib/msro/openaireplus/workflows/index/openaireLayoutToRecordStylesheet.xsl</PARAM>
|
64 |
|
|
<PARAM name="oafSchemaLocationProperty" type="string" managedBy="system" required="true">oaf.schema.location</PARAM>
|
65 |
26600
|
sandro.lab
|
</PARAMETERS>
|
66 |
|
|
<ARCS>
|
67 |
|
|
<ARC to="cleanupXml" />
|
68 |
|
|
<ARC to="cleanupRotten" />
|
69 |
|
|
</ARCS>
|
70 |
|
|
</NODE>
|
71 |
|
|
<NODE name="cleanupXml" type="DeleteHdfsPathJob">
|
72 |
|
|
<DESCRIPTION>hdfs cleanup (xml)</DESCRIPTION>
|
73 |
|
|
<PARAMETERS>
|
74 |
|
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
75 |
|
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
76 |
|
|
{
|
77 |
|
|
'path' : 'outputRecordsPath'
|
78 |
|
|
}
|
79 |
|
|
</PARAM>
|
80 |
|
|
</PARAMETERS>
|
81 |
|
|
<ARCS>
|
82 |
|
|
<ARC to="group" />
|
83 |
|
|
</ARCS>
|
84 |
|
|
</NODE>
|
85 |
|
|
<NODE name="cleanupRotten" type="DeleteHdfsPathJob">
|
86 |
|
|
<DESCRIPTION>hdfs cleanup (rotten)</DESCRIPTION>
|
87 |
|
|
<PARAMETERS>
|
88 |
|
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
89 |
|
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
90 |
|
|
{
|
91 |
|
|
'path' : 'rottenRecordsPath'
|
92 |
|
|
}
|
93 |
|
|
</PARAM>
|
94 |
|
|
</PARAMETERS>
|
95 |
|
|
<ARCS>
|
96 |
|
|
<ARC to="group" />
|
97 |
|
|
</ARCS>
|
98 |
|
|
</NODE>
|
99 |
|
|
<NODE name="group" type="SubmitHadoopJob" isJoin="true">
|
100 |
|
|
<DESCRIPTION>M/R group entities</DESCRIPTION>
|
101 |
|
|
<PARAMETERS>
|
102 |
|
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
103 |
|
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">prepareIndexDataJob</PARAM>
|
104 |
|
|
<PARAM required="true" type="string" name="sysParams" managedBy="system">
|
105 |
|
|
{
|
106 |
|
|
'hbase.mapred.inputtable' : 'hbase.mapred.datatable',
|
107 |
|
|
'hbase.mapreduce.inputtable' : 'hbase.mapred.datatable'
|
108 |
|
|
}
|
109 |
|
|
</PARAM>
|
110 |
|
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
111 |
|
|
{
|
112 |
|
|
'mapred.output.dir' : 'outputRecordsPath',
|
113 |
|
|
'index.entity.links' : 'index.entity.links',
|
114 |
28835
|
claudio.at
|
'oaf.schema.location' : 'oaf.schema.location',
|
115 |
28824
|
claudio.at
|
'contextmap' : 'contextmap',
|
116 |
|
|
'relClasses' : 'relClasses'
|
117 |
26600
|
sandro.lab
|
}
|
118 |
|
|
</PARAM>
|
119 |
|
|
</PARAMETERS>
|
120 |
|
|
<ARCS>
|
121 |
|
|
<ARC to="updateIndex" />
|
122 |
|
|
</ARCS>
|
123 |
|
|
</NODE>
|
124 |
|
|
<NODE name="updateIndex" type="SubmitHadoopJob">
|
125 |
|
|
<DESCRIPTION>M/O index records</DESCRIPTION>
|
126 |
|
|
<PARAMETERS>
|
127 |
|
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
128 |
|
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">indexFeedJob</PARAM>
|
129 |
|
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
130 |
|
|
{
|
131 |
|
|
'mapred.input.dir' : 'outputRecordsPath',
|
132 |
|
|
'mapred.output.dir' : 'rottenRecordsPath',
|
133 |
|
|
'index.xslt' : 'index.xslt',
|
134 |
30031
|
claudio.at
|
'index.solr.url' : 'index.solr.url',
|
135 |
26600
|
sandro.lab
|
'index.solr.collection' : 'index.solr.collection',
|
136 |
|
|
'index.buffer.flush.threshold' : 'index.buffer.flush.threshold',
|
137 |
|
|
'index.shutdown.wait.time' : 'index.shutdown.wait.time',
|
138 |
|
|
'index.solr.sim.mode' : 'index.solr.sim.mode',
|
139 |
|
|
'index.feed.timestamp' : 'index.feed.timestamp'
|
140 |
|
|
}
|
141 |
|
|
</PARAM>
|
142 |
|
|
</PARAMETERS>
|
143 |
|
|
<ARCS>
|
144 |
30021
|
claudio.at
|
<ARC to="finalize" />
|
145 |
28824
|
claudio.at
|
</ARCS>
|
146 |
|
|
</NODE>
|
147 |
26600
|
sandro.lab
|
<NODE name="finalize" type="FinalizeIndexFeeding">
|
148 |
|
|
<DESCRIPTION>commit changes</DESCRIPTION>
|
149 |
|
|
<PARAMETERS />
|
150 |
|
|
<ARCS>
|
151 |
|
|
<ARC to="updateDs" />
|
152 |
30021
|
claudio.at
|
<ARC to="findSearchService" />
|
153 |
26600
|
sandro.lab
|
</ARCS>
|
154 |
|
|
</NODE>
|
155 |
|
|
<NODE name="updateDs" type="IndexDsUpdateJob">
|
156 |
|
|
<DESCRIPTION>update DS</DESCRIPTION>
|
157 |
|
|
<PARAMETERS />
|
158 |
|
|
<ARCS>
|
159 |
|
|
<ARC to="success" />
|
160 |
|
|
</ARCS>
|
161 |
|
|
</NODE>
|
162 |
28824
|
claudio.at
|
<NODE name="findSearchService" type="FindSearchService">
|
163 |
|
|
<DESCRIPTION>find search service</DESCRIPTION>
|
164 |
30781
|
claudio.at
|
<PARAMETERS>
|
165 |
|
|
<PARAM name="xquery" type="string" managedBy="user" required="false">/RESOURCE_PROFILE[.//RESOURCE_TYPE/@value='SearchServiceResourceType' and .//RESOURCE_KIND/@value='ServiceResources']/HEADER/RESOURCE_IDENTIFIER/@value/string()</PARAM>
|
166 |
|
|
<PARAM name="xqueryForServiceIdParam" type="string" managedBy="system" required="true">xqueryForSearchService</PARAM>
|
167 |
|
|
</PARAMETERS>
|
168 |
28824
|
claudio.at
|
<ARCS>
|
169 |
|
|
<ARC name="found" to="switch" />
|
170 |
|
|
<ARC name="notFound" to="success" />
|
171 |
|
|
</ARCS>
|
172 |
|
|
</NODE>
|
173 |
|
|
<NODE name="switch" type="SwitchIndex">
|
174 |
|
|
<DESCRIPTION>switch index</DESCRIPTION>
|
175 |
30781
|
claudio.at
|
<PARAMETERS>
|
176 |
|
|
<PARAM name="xqueryForServiceIdParam" type="string" managedBy="system" required="true">xqueryForSearchService</PARAM>
|
177 |
|
|
<PARAM name="inputIndexIdParam" type="string" managedBy="system" required="true">index_id</PARAM>
|
178 |
|
|
<PARAM name="outputIndexIdParam" type="string" managedBy="system" required="true">IndexId</PARAM>
|
179 |
|
|
</PARAMETERS>
|
180 |
28824
|
claudio.at
|
<ARCS>
|
181 |
|
|
<ARC to="success" />
|
182 |
|
|
</ARCS>
|
183 |
|
|
</NODE>
|
184 |
26600
|
sandro.lab
|
</CONFIGURATION>
|
185 |
|
|
<STATUS />
|
186 |
|
|
</BODY>
|
187 |
|
|
</RESOURCE_PROFILE>
|
188 |
|
|
|