1
|
<RESOURCE_PROFILE>
|
2
|
<HEADER>
|
3
|
<RESOURCE_IDENTIFIER value="0631da9f-8a4f-4412-828b-37f7a0aa6d71_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
|
4
|
<RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
|
5
|
<RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
|
6
|
<RESOURCE_URI value=""/>
|
7
|
<DATE_OF_CREATION value="2001-12-31T12:00:00"/>
|
8
|
</HEADER>
|
9
|
<BODY>
|
10
|
<HADOOP_JOB name="brokerEnrichmentSoftwareLinksJob" type="mapreduce">
|
11
|
<DESCRIPTION>map reduce job that calculates the enrichment events based on the publications dedup results</DESCRIPTION>
|
12
|
|
13
|
<STATIC_CONFIGURATION><!-- I/O FORMAT -->
|
14
|
<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat"/>
|
15
|
<PROPERTY key="mapreduce.outputformat.class" value="org.elasticsearch.hadoop.mr.EsOutputFormat"/>
|
16
|
|
17
|
<!-- MAPPER -->
|
18
|
<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.broker.enrich.SoftwareEnrichmentMapper"/>
|
19
|
<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable"/>
|
20
|
<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable"/>
|
21
|
|
22
|
<!-- REDUCER -->
|
23
|
<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.broker.enrich.SoftwareEnrichmentReducer"/>
|
24
|
<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text"/>
|
25
|
<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text"/>
|
26
|
|
27
|
<!-- MISC -->
|
28
|
<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false"/>
|
29
|
<PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/>
|
30
|
<PROPERTY key="mapreduce.map.speculative" value="false"/>
|
31
|
<PROPERTY key="mapreduce.reduce.speculative" value="false"/>
|
32
|
<PROPERTY key="dfs.blocksize" value="256M"/>
|
33
|
<PROPERTY key="mapred.reduce.tasks" value="4"/>
|
34
|
|
35
|
<PROPERTY key="broker.baseurl.publication" value="https://explore.openaire.eu/search/publication?articleId=%s"/>
|
36
|
<PROPERTY key="broker.baseurl.dataset" value="https://explore.openaire.eu/search/dataset?datasetId=%s"/>
|
37
|
<PROPERTY key="broker.baseurl.software" value="https://explore.openaire.eu/search/software?softwareId=%s"/>
|
38
|
<PROPERTY key="broker.baseurl.other" value="https://explore.openaire.eu/search/other?orpId=%s"/>
|
39
|
|
40
|
|
41
|
<!-- ES -->
|
42
|
<PROPERTY key="es.nodes" value="ip-90-147-167-137.ct1.garrservices.it:9200,ip-90-147-167-126.ct1.garrservices.it:9200,ip-90-147-167-13.ct1.garrservices.it:9200,ip-90-147-167-125.ct1.garrservices.it:9200"/>
|
43
|
<PROPERTY key="es.nodes.resolve.hostname" value="false"/>
|
44
|
<PROPERTY key="es.batch.write.retry.count " value="10"/>
|
45
|
<PROPERTY key="es.batch.size.entries " value="500"/>
|
46
|
<PROPERTY key="es.nodes.wan.only" value="true"/>
|
47
|
<PROPERTY key="es.resource" value="events_{infra}/event"/>
|
48
|
<PROPERTY key="es.input.json" value="yes"/>
|
49
|
<PROPERTY key="es.mapping.id" value="eventId"/>
|
50
|
|
51
|
<!-- BROKER -->
|
52
|
<PROPERTY key="broker.datasource.id.whitelist" value=""/>
|
53
|
<PROPERTY key="broker.datasource.id.blacklist" value=""/>
|
54
|
<PROPERTY key="broker.datasource.untrusted.oa.list" value="opendoar____::8b6dd7db9af49e67306feb59a8bdc52c"/>
|
55
|
<PROPERTY key="broker.datasource.type.whitelist" value="pubsrepository::unknown,pubsrepository::institutional,pubsrepository::thematic"/>
|
56
|
|
57
|
<!-- <PROPERTY key="user.name" value="dnet" /> -->
|
58
|
<!-- Uncomment to override the default lib path -->
|
59
|
<!-- <PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
|
60
|
</STATIC_CONFIGURATION>
|
61
|
<JOB_INTERFACE>
|
62
|
<PARAM description="source hbase table" name="hbase.mapred.inputtable" required="true"/>
|
63
|
<PARAM description="source hbase table" name="hbase.mapreduce.inputtable" required="true"/>
|
64
|
</JOB_INTERFACE>
|
65
|
<SCAN>
|
66
|
<FILTERS operator="MUST_PASS_ONE">
|
67
|
<FILTER type="prefix" value="40" />
|
68
|
<FILTER type="prefix" value="50" />
|
69
|
</FILTERS>
|
70
|
<FAMILIES>
|
71
|
<FAMILY value="result" />
|
72
|
<FAMILY value="resultResult_relationship_isRelatedTo" />
|
73
|
</FAMILIES>
|
74
|
</SCAN>
|
75
|
</HADOOP_JOB>
|
76
|
<STATUS>
|
77
|
<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
|
78
|
<RUNNING_INSTANCES value="0"/>
|
79
|
<CUMULATIVE_RUN value="0" />
|
80
|
</STATUS>
|
81
|
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
|
82
|
</BODY>
|
83
|
</RESOURCE_PROFILE>
|