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="5efd62d9-b7eb-4d62-962c-be15bd1e9ded_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>Generate Stats</WORKFLOW_NAME>
|
12
|
<WORKFLOW_TYPE>Provision</WORKFLOW_TYPE>
|
13
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
14
|
<CONFIGURATION start="manual">
|
15
|
<NODE name="fetchEntityLinks" type="LoadEntityLinksJob" isStart="true">
|
16
|
<DESCRIPTION />
|
17
|
<PARAMETERS/>
|
18
|
<ARCS>
|
19
|
<ARC to="updateStats" />
|
20
|
</ARCS>
|
21
|
</NODE>
|
22
|
<NODE name="prepareParams" type="PrepareStatsParams" isStart="true">
|
23
|
<DESCRIPTION />
|
24
|
<PARAMETERS>
|
25
|
<PARAM required="true" type="string" name="paramsMapJson" managedBy="user">
|
26
|
{
|
27
|
'oozieWfApplicationPath' : '/user/eri.katsari/stats/oozie_app',
|
28
|
'statsDbUrl' : 'jdbc:postgresql://node1.t.openaire.research-infrastructures.eu:5432/stats',
|
29
|
'statsDbUser' : 'sqoop',
|
30
|
'statsDbPass' : 'sqoop',
|
31
|
'statsDbDriver' : 'org.postgresql.Driver',
|
32
|
'statsDbTableMap' : 'datasourceLanguage=datasource_languages,datasource=datasource,project=project,result=result,organization=organization,datasourceOrganization=datasource_organizations,datasourceTopic=datasource_topics,projectOrganization=project_organizations,resultClaim=result_claims,resultClassification=result_classifications,resultConcept=result_concepts,resultLanguage=result_languages,resultOrganization=result_organizations,resultResult=result_results,resultProject=project_results,resultResult=result_results,resultTopic=result_topics,category=category,claim=claim,concept=concept,datasourceLanguage=datasource_languages,resultLanguage=result_languages,resultDatasource=result_datasources',
|
33
|
'statsSqoopRecsPerStatement' : '1000',
|
34
|
'statsSqoopStatementPerTrans' : '1000',
|
35
|
'statsSqoopReducersCount' : '4',
|
36
|
|
37
|
'statsOutputPath' : '/tmp/test_stats/',
|
38
|
'statsNullStringField' : 'NULL',
|
39
|
'statsNullNumericField' : '-1',
|
40
|
'statsEnclosingCharacter' : '*',
|
41
|
'statsDelimCharacter' : '!',
|
42
|
|
43
|
'out1' : 'datasource',
|
44
|
'out2' : 'project',
|
45
|
'out3' : 'organization',
|
46
|
'out4' : 'datasourceOrganization',
|
47
|
'out5' : 'datasourceTopic',
|
48
|
'out6' : 'datasourceLanguage',
|
49
|
'out7' : 'projectOrganization',
|
50
|
'out8' : 'resultClaim',
|
51
|
'out9' : 'resultClassification',
|
52
|
'out10' : 'resultConcept',
|
53
|
'out11' : 'resultLanguage',
|
54
|
'out12' : 'resultOrganization',
|
55
|
'out13' : 'resultResult',
|
56
|
'out14' : 'resultProject',
|
57
|
'out15' : 'category',
|
58
|
'out16' : 'resultTopic',
|
59
|
'out17' : 'resultDatasource',
|
60
|
'out18' : 'result',
|
61
|
'out19' : 'claim',
|
62
|
'out20' : 'concept'
|
63
|
}
|
64
|
</PARAM>
|
65
|
</PARAMETERS>
|
66
|
<ARCS>
|
67
|
<ARC to="updateStats" />
|
68
|
</ARCS>
|
69
|
</NODE>
|
70
|
<NODE name="updateStats" type="SubmitHadoopJob" isJoin="true">
|
71
|
<DESCRIPTION>Update stats DB</DESCRIPTION>
|
72
|
<PARAMETERS>
|
73
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
74
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">StatsExportJob</PARAM>
|
75
|
<PARAM required="true" type="string" name="sysParams" managedBy="system">
|
76
|
{
|
77
|
'Stats_Hbase_Source_Table' : 'hbase.mapred.datatable'
|
78
|
}
|
79
|
</PARAM>
|
80
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
81
|
{
|
82
|
'nameNode' : 'nameNode',
|
83
|
'jobTracker' : 'jobTracker',
|
84
|
'oozie.wf.application.path' : 'oozieWfApplicationPath',
|
85
|
'Stats_db_Url' : 'statsDbUrl',
|
86
|
'Stats_db_User' : 'statsDbUser',
|
87
|
'Stats_db_Pass' : 'statsDbPass',
|
88
|
'Stats_db_Driver' : 'statsDbDriver',
|
89
|
'Stats_db_table_map' : 'statsDbTableMap',
|
90
|
'Stats_sqoop_RecsPerStatement' : 'statsSqoopRecsPerStatement',
|
91
|
'Stats_sqoop_StatementPerTrans' : 'statsSqoopStatementPerTrans',
|
92
|
'Stats_sqoop_ReducersCount' : 'statsSqoopReducersCount',
|
93
|
'Stats_output_Path' : 'statsOutputPath',
|
94
|
'Stats_null_String_Field' : 'statsNullStringField',
|
95
|
'Stats_null_Numeric_Field' : 'statsNullNumericField',
|
96
|
'Stats_enclosing_Character' : 'statsEnclosingCharacter',
|
97
|
'Stats_delim_Character' : 'statsDelimCharacter',
|
98
|
'out1' : 'out1',
|
99
|
'out2' : 'out2',
|
100
|
'out3' : 'out3',
|
101
|
'out4' : 'out4',
|
102
|
'out5' : 'out5',
|
103
|
'out6' : 'out6',
|
104
|
'out7' : 'out7',
|
105
|
'out8' : 'out8',
|
106
|
'out9' : 'out9',
|
107
|
'out10' : 'out10',
|
108
|
'out11' : 'out11',
|
109
|
'out12' : 'out12',
|
110
|
'out13' : 'out13',
|
111
|
'out14' : 'out14',
|
112
|
'out15' : 'out15',
|
113
|
'out16' : 'out16',
|
114
|
'out17' : 'out17',
|
115
|
'out18' : 'out18',
|
116
|
'out19' : 'out19',
|
117
|
'out20' : 'out20',
|
118
|
'isLookupEndpoint' : 'isLookupEndpoint',
|
119
|
'Stats_indexConf' : 'index.entity.links'
|
120
|
}
|
121
|
</PARAM>
|
122
|
</PARAMETERS>
|
123
|
<ARCS>
|
124
|
<ARC to="success" />
|
125
|
</ARCS>
|
126
|
</NODE>
|
127
|
</CONFIGURATION>
|
128
|
<STATUS />
|
129
|
</BODY>
|
130
|
</RESOURCE_PROFILE>
|
131
|
|
132
|
|