|
1
|
<RESOURCE_PROFILE>
|
|
2
|
<HEADER>
|
|
3
|
<RESOURCE_IDENTIFIER value="1046e131-fc27-47df-898e-8ce396c7647c_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
|
|
4
|
<RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
|
|
5
|
<RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
|
|
6
|
<RESOURCE_URI value=""/>
|
|
7
|
<DATE_OF_CREATION value="2018-09-10T18:00:00"/>
|
|
8
|
</HEADER>
|
|
9
|
<BODY>
|
|
10
|
<HADOOP_JOB name="propagationCountryFromInstitutionalRepositories" type="mapreduce">
|
|
11
|
<DESCRIPTION>map reduce job that propagates country value of the organization to products belonging to institutional repositories related to the organization (pubsrepository:institutional)</DESCRIPTION>
|
|
12
|
<STATIC_CONFIGURATION><!-- I/O FORMAT -->
|
|
13
|
<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat"/>
|
|
14
|
<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat"/>
|
|
15
|
|
|
16
|
<!-- MAPPER -->
|
|
17
|
<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.propagation.country.institutionalrepositories.PropagationCountryFromDsOrgResultMapper"/>
|
|
18
|
<PROPERTY key="mapred.mapoutput.key.class" value="eu.dnetlib.data.mapreduce.hbase.propagation.compositekeys.InstOrgKey"/>
|
|
19
|
<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text"/>
|
|
20
|
|
|
21
|
<!-- REDUCER -->
|
|
22
|
<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.propagation.country.institutionalrepositories.PropagationCountryFromDsOrgResultReducer"/>
|
|
23
|
<PROPERTY key="mapred.output.key.class" value="eu.dnetlib.data.mapreduce.hbase.propagation.compositekeys.InstOrgKey"/>
|
|
24
|
<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text"/>
|
|
25
|
|
|
26
|
<!-- PARTITIONER -->
|
|
27
|
<PROPERTY key="mapred.partitioner.class" value="eu.dnetlib.data.mapreduce.hbase.propagation.compositekeys.NaturalInstOrgKeyPartitioner"/>
|
|
28
|
<PROPERTY key="mapreduce.partitioner.class" value="eu.dnetlib.data.mapreduce.hbase.propagation.compositekeys.NaturalInstOrgKeyPartitioner"/>
|
|
29
|
<PROPERTY key="mapred.output.value.groupfn.class" value="eu.dnetlib.data.mapreduce.hbase.propagation.compositekeys.NaturalInstOrgKeyGroupingComparator"/>
|
|
30
|
<PROPERTY key="mapreduce.output.value.groupfn.class" value="eu.dnetlib.data.mapreduce.hbase.propagation.compositekeys.NaturalInstOrgKeyGroupingComparator"/>
|
|
31
|
|
|
32
|
<!-- MISC -->
|
|
33
|
<PROPERTY key="mapred.compress.map.output" value="true"/>
|
|
34
|
<PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/>
|
|
35
|
<PROPERTY key="mapreduce.map.speculative" value="false"/>
|
|
36
|
<PROPERTY key="mapreduce.reduce.speculative" value="false"/>
|
|
37
|
<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false"/>
|
|
38
|
<PROPERTY key="mapred.reduce.tasks" value="20"/>
|
|
39
|
|
|
40
|
<!-- CSV of datasource typologies considered in the processing -->
|
|
41
|
<PROPERTY key="datasource.types" value="pubsrepository::institutional"/>
|
|
42
|
|
|
43
|
<!-- <PROPERTY key="user.name" value="dnet" /> --><!-- Uncomment to override the default lib path -->
|
|
44
|
<PROPERTY key="job.lib" value="/user/dnet/openaire/dnet-mapreduce-jobs-assembly-country_propagation-1.1.1-SNAPSHOT.jar"/>
|
|
45
|
</STATIC_CONFIGURATION>
|
|
46
|
<JOB_INTERFACE>
|
|
47
|
<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table"/>
|
|
48
|
<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table"/>
|
|
49
|
<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table"/>
|
|
50
|
<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table"/><!--<PARAM name="countryPropagation.conf" required="true" description="configuration for country propagation" />-->
|
|
51
|
</JOB_INTERFACE>
|
|
52
|
<SCAN>
|
|
53
|
<FILTERS/>
|
|
54
|
<FAMILIES/>
|
|
55
|
</SCAN>
|
|
56
|
</HADOOP_JOB>
|
|
57
|
<STATUS>
|
|
58
|
<LAST_SUBMISSION_DATE value="2018-09-18T12:13:07+02:00"/>
|
|
59
|
<RUNNING_INSTANCES value="1"/>
|
|
60
|
<CUMULATIVE_RUN value="30"/>
|
|
61
|
</STATUS>
|
|
62
|
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
|
|
63
|
</BODY>
|
|
64
|
</RESOURCE_PROFILE>
|