|
1
|
<?xml version="1.0" encoding="UTF-8"?>
|
|
2
|
<RESOURCE_PROFILE>
|
|
3
|
<HEADER>
|
|
4
|
<RESOURCE_IDENTIFIER value="96c45851-28ed-4cd3-8942-bbb350e4dfec_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
|
|
5
|
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
|
|
6
|
<RESOURCE_KIND value="WorkflowDSResources"/>
|
|
7
|
<RESOURCE_URI value=""/>
|
|
8
|
<DATE_OF_CREATION value="2014-08-01T18:13:51.0Z"/>
|
|
9
|
</HEADER>
|
|
10
|
<BODY>
|
|
11
|
<WORKFLOW_NAME>OAI Store feed</WORKFLOW_NAME>
|
|
12
|
<WORKFLOW_TYPE>Provision</WORKFLOW_TYPE>
|
|
13
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
|
14
|
<CONFIGURATION start="manual">
|
|
15
|
<NODE name="setHdfs" type="SetHdfsFile" isStart="true">
|
|
16
|
<DESCRIPTION/>
|
|
17
|
<PARAMETERS>
|
|
18
|
<PARAM name="hdfsPath" type="string" managedBy="user" required="true">/tmp/indexrecords_db_openaireplus_node6_t_DMF.seq</PARAM>
|
|
19
|
<PARAM name="hdfsPathParam" type="string" managedBy="system" required="true">inputRecordsPath</PARAM>
|
|
20
|
</PARAMETERS>
|
|
21
|
<ARCS>
|
|
22
|
<ARC to="prepareOAI"/>
|
|
23
|
</ARCS>
|
|
24
|
</NODE>
|
|
25
|
|
|
26
|
<NODE name="setSkipDuplicates" type="SetEnvParameter" isStart="true">
|
|
27
|
<DESCRIPTION/>
|
|
28
|
<PARAMETERS>
|
|
29
|
<PARAM name="parameterName" type="string" managedBy="system" required="true">skipDuplicates</PARAM>
|
|
30
|
<PARAM name="parameterValue" type="string" managedBy="system" required="true">true</PARAM>
|
|
31
|
</PARAMETERS>
|
|
32
|
<ARCS>
|
|
33
|
<ARC to="prepareOAI"/>
|
|
34
|
</ARCS>
|
|
35
|
</NODE>
|
|
36
|
|
|
37
|
<NODE name="setDuplicateXPath" type="SetEnvParameter" isStart="true">
|
|
38
|
<DESCRIPTION/>
|
|
39
|
<PARAMETERS>
|
|
40
|
<PARAM name="parameterName" type="string" managedBy="system" required="true">duplicateXPath</PARAM>
|
|
41
|
<PARAM name="parameterValue" type="string" managedBy="system" required="true">//*[local-name()='entity']//*[local-name()='datainfo']/*[local-name()='deletedbyinference'][./text() = 'true']</PARAM>
|
|
42
|
</PARAMETERS>
|
|
43
|
<ARCS>
|
|
44
|
<ARC to="prepareOAI"/>
|
|
45
|
</ARCS>
|
|
46
|
</NODE>
|
|
47
|
|
|
48
|
<NODE name="prepareOAI" type="PrepareOaiJob" isJoin="true">
|
|
49
|
<DESCRIPTION>Prepare oai feeding</DESCRIPTION>
|
|
50
|
<PARAMETERS>
|
|
51
|
<PARAM name="oaiDbName" type="string" managedBy="user" required="true">oaistore</PARAM>
|
|
52
|
<PARAM managedBy="user" name="oaiFormat" required="true" type="string">oaf</PARAM>
|
|
53
|
<PARAM managedBy="user" name="oaiLayout" required="true" type="string">index</PARAM>
|
|
54
|
<PARAM managedBy="user" name="oaiInterpretation" required="true" type="string">openaire</PARAM>
|
|
55
|
</PARAMETERS>
|
|
56
|
<ARCS>
|
|
57
|
<ARC to="OAIDropStore"/>
|
|
58
|
</ARCS>
|
|
59
|
</NODE>
|
|
60
|
<NODE name="OAIDropStore" type="OAIDropStore">
|
|
61
|
<DESCRIPTION>Drop the store: upserts are too expensive!</DESCRIPTION>
|
|
62
|
<PARAMETERS/>
|
|
63
|
<ARCS>
|
|
64
|
<ARC to="OAICreateStore"/>
|
|
65
|
</ARCS>
|
|
66
|
</NODE>
|
|
67
|
<NODE name="OAICreateStore" type="OAICreateStore">
|
|
68
|
<DESCRIPTION>Create the OAI store</DESCRIPTION>
|
|
69
|
<PARAMETERS/>
|
|
70
|
<ARCS>
|
|
71
|
<ARC to="feedOAI"/>
|
|
72
|
</ARCS>
|
|
73
|
</NODE>
|
|
74
|
|
|
75
|
<NODE name="feedOAI" type="SubmitHadoopJob">
|
|
76
|
<DESCRIPTION>M/O oai feeding</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">oaiFeedJob</PARAM>
|
|
80
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
|
81
|
{
|
|
82
|
'mapred.input.dir' : 'inputRecordsPath',
|
|
83
|
'services.publisher.oai.collection' : 'oai_collection',
|
|
84
|
'oaiConfiguration' : 'oai_configuration',
|
|
85
|
'oai.feed.date' : 'oai.feed.date',
|
|
86
|
'services.publisher.oai.db' : 'oai_dbName',
|
|
87
|
'services.publisher.oai.skipDuplicates':'skipDuplicates',
|
|
88
|
'services.publisher.oai.duplicateXPath':'duplicateXPath'
|
|
89
|
}
|
|
90
|
</PARAM>
|
|
91
|
<PARAM required="true" type="string" name="sysParams" managedBy="system">
|
|
92
|
{
|
|
93
|
'services.publisher.oai.host' : 'services.publisher.oai.host',
|
|
94
|
'services.publisher.oai.port' : 'services.publisher.oai.port'
|
|
95
|
}
|
|
96
|
</PARAM>
|
|
97
|
</PARAMETERS>
|
|
98
|
<ARCS>
|
|
99
|
<ARC to="success"/>
|
|
100
|
</ARCS>
|
|
101
|
</NODE>
|
|
102
|
|
|
103
|
</CONFIGURATION>
|
|
104
|
<STATUS/>
|
|
105
|
</BODY>
|
|
106
|
</RESOURCE_PROFILE>
|