1
|
<RESOURCE_PROFILE>
|
2
|
<HEADER>
|
3
|
<RESOURCE_IDENTIFIER value="bf0ab07b-36bf-4164-ab73-342bfb11e51a_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
|
4
|
<RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
|
5
|
<RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
|
6
|
<RESOURCE_URI value=""/>
|
7
|
<DATE_OF_CREATION value="2016-05-23T10:24:25+00:00"/>
|
8
|
</HEADER>
|
9
|
<BODY>
|
10
|
<HADOOP_JOB name="LodGeneration" type="oozie">
|
11
|
<DESCRIPTION>Job for importing data from HBASE to the relational Stats Database</DESCRIPTION>
|
12
|
<STATIC_CONFIGURATION>
|
13
|
|
14
|
<!-- Cluster wide -->
|
15
|
<PROPERTY key="queueName" value="default"/>
|
16
|
<PROPERTY key="user.name" value="dnet"/> <!-- username = sqoop?? -->
|
17
|
<PROPERTY key="workingDir" value="/lib/lod/releases/working_dir"/>
|
18
|
|
19
|
<!--Reducers -->
|
20
|
<PROPERTY key="numReducers" value="4"/>
|
21
|
|
22
|
<!-- Path -->
|
23
|
<PROPERTY key="oozie.wf.application.path" value="hdfs://dm-cluster-nn/lib/stats/releases/stats/oozie_app"/><!-- edit this property! -->
|
24
|
|
25
|
<!-- DB props-->
|
26
|
<PROPERTY key="Stats_db_Url" value="jdbc:postgresql://stats.openaire.eu:5432/stats"/>
|
27
|
<PROPERTY key="Stats_db_User" value="sqoop"/>
|
28
|
<PROPERTY key="Stats_db_Pass" value="sqoop"/>
|
29
|
<PROPERTY key="Stats_db_Driver" value="org.postgresql.Driver"/>
|
30
|
|
31
|
<!--Sqoop props-->
|
32
|
<PROPERTY key="Stats_sqoop_RecsPerStatement" value="10000"/>
|
33
|
<PROPERTY key="Stats_sqoop_StatementPerTrans" value="1000000"/>
|
34
|
<PROPERTY key="Stats_sqoop_ReducersCount" value="8"/>
|
35
|
<PROPERTY key="Stats_output_Path" value="/tmp/stats/"/>
|
36
|
|
37
|
<!--Parser props-->
|
38
|
<PROPERTY key="Stats_db_table_map" value="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,category=category,context=context,datasourceLanguage=datasource_languages,resultLanguage=result_languages,resultDatasource=result_datasources, person=person,personResult=person_results,resultDescription=result_descriptions,datasourceStats=datasource_stats,organizationStats=organization_stats,projectStats=project_stats,resultStats=result_stats,resultOid=result_oids,projectOid=project_oids,personOid=person_oids,datasourceOid=datasource_oids,organizationOid=organization_oids,resultPid=result_pids"/>
|
39
|
<PROPERTY key="Stats_null_String_Field" value="null"/>
|
40
|
<PROPERTY key="Stats_null_Numeric_Field" value="null"/>
|
41
|
<PROPERTY key="Stats_enclosing_Character" value="#"/>
|
42
|
<PROPERTY key="Stats_delim_Character" value="!"/>
|
43
|
<PROPERTY key="Stats_getdeletedbyinference" value="false"/>
|
44
|
|
45
|
<!--outputs -->
|
46
|
<PROPERTY key="out1" value="datasource"/>
|
47
|
<PROPERTY key="out2" value="project"/>
|
48
|
<PROPERTY key="out3" value="organization"/>
|
49
|
<PROPERTY key="out4" value="datasourceOrganization"/>
|
50
|
<PROPERTY key="out5" value="datasourceTopic"/>
|
51
|
<PROPERTY key="out6" value="datasourceLanguage"/>
|
52
|
<PROPERTY key="out7" value="projectOrganization"/>
|
53
|
<PROPERTY key="out8" value="resultClaim"/>
|
54
|
<PROPERTY key="out9" value="resultClassification"/>
|
55
|
<PROPERTY key="out10" value="resultConcept"/>
|
56
|
<PROPERTY key="out11" value="resultLanguage"/>
|
57
|
<PROPERTY key="out12" value="resultOrganization"/>
|
58
|
<PROPERTY key="out13" value="resultResult"/>
|
59
|
<PROPERTY key="out14" value="resultProject"/>
|
60
|
<PROPERTY key="out15" value="resultTopic"/>
|
61
|
<PROPERTY key="out16" value="resultDatasource"/>
|
62
|
<PROPERTY key="out17" value="result"/>
|
63
|
<PROPERTY key="out18" value="context"/>
|
64
|
<PROPERTY key="out19" value="concept"/>
|
65
|
<PROPERTY key="out20" value="category"/>
|
66
|
<PROPERTY key="out21" value="person"/>
|
67
|
<PROPERTY key="out22" value="personResult"/>
|
68
|
<PROPERTY key="out23" value="resultCitation"/>
|
69
|
<PROPERTY key="out24" value="resultPid"/>
|
70
|
<PROPERTY key="out25" value="resultOid"/>
|
71
|
<PROPERTY key="out26" value="projectOid"/>
|
72
|
<PROPERTY key="out27" value="projectKeyword"/>
|
73
|
<PROPERTY key="out28" value="projectSubject"/>
|
74
|
<PROPERTY key="out29" value="organizationOid"/>
|
75
|
<PROPERTY key="out30" value="datasourceOid"/>
|
76
|
<PROPERTY key="out31" value="personOid"/>
|
77
|
<PROPERTY key="out32" value="projectPerson"/>
|
78
|
<PROPERTY key="out33" value="datasourceStats"/>
|
79
|
<PROPERTY key="out34" value="projectStats"/>
|
80
|
<PROPERTY key="out35" value="organizationStats"/>
|
81
|
<PROPERTY key="out36" value="resultStats"/>
|
82
|
<PROPERTY key="out37" value="person"/>
|
83
|
<PROPERTY key="out38" value="resultDescription"/>
|
84
|
<PROPERTY key="out39" value="organizationOrganization"/>
|
85
|
|
86
|
<!--usage stats additions-->
|
87
|
<PROPERTY key="piwik_username" value="uminho-pilot"/>
|
88
|
<PROPERTY key="piwik_pass" value="eeTechoo4eeh"/>
|
89
|
<PROPERTY key="piwik_httpProtocol" value="https"/>
|
90
|
<PROPERTY key="piwik_url" value="analytics.openaire.eu"/>
|
91
|
<PROPERTY key="piwik_siteId" value="5"/>
|
92
|
<PROPERTY key="piwik_startDate" value="2015-01-01"/>
|
93
|
<PROPERTY key="piwik_finalDate" value="2016-01-03"/>
|
94
|
<PROPERTY key="piwik_logsPath" value="/tmp/piwikLogs/"/>
|
95
|
<PROPERTY key="piwik_filterOffset" value="5"/>
|
96
|
<PROPERTY key="piwiki_schema" value="usagestats"/>
|
97
|
</STATIC_CONFIGURATION>
|
98
|
<JOB_INTERFACE>
|
99
|
<PARAM description="hdfs name node" name="nameNode" required="true"/>
|
100
|
<PARAM description="job tracker name" name="jobTracker" required="true"/>
|
101
|
<PARAM description="Hbase Table with Protobuffs." name="Stats_Hbase_Source_Table" required="true"/>
|
102
|
<PARAM description="Index Entity Links configuration." name="Stats_indexConf" required="true"/>
|
103
|
<PARAM description="IS lookup service endpoint" name="isLookupEndpoint" required="true"/>
|
104
|
</JOB_INTERFACE>
|
105
|
</HADOOP_JOB>
|
106
|
<STATUS>
|
107
|
<LAST_SUBMISSION_DATE value="2016-05-23T10:43:52+00:00"/>
|
108
|
<RUNNING_INSTANCES value="0"/>
|
109
|
<CUMULATIVE_RUN value="142"/>
|
110
|
</STATUS>
|
111
|
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
|
112
|
</BODY>
|
113
|
</RESOURCE_PROFILE>
|