Project

General

Profile

1
<RESOURCE_PROFILE>
2
	<HEADER>
3
		<RESOURCE_IDENTIFIER value="3f544a36-f123-4f5c-acf4-7c25f6591ec4_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="personCsvJoinJob" type="mapreduce">
11
			<DESCRIPTION>map reduce job that joins person entities by "surname+first name letter" and serialise the output as csv</DESCRIPTION>
12
			<STATIC_CONFIGURATION>
13

    
14
				<!-- I/O FORMAT -->
15
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat" />
16
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.LazyOutputFormat" />
17
				<PROPERTY key="mapreduce.output.lazyoutputformat.outputformat" value="org.apache.hadoop.mapreduce.lib.output.TextOutputFormat" />
18

    
19
				<!-- MAPPER -->
20
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.experiment.JoinPersonGroupMapper" />
21
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
22
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text" />
23

    
24
				<!-- REDUCER -->
25
				<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.experiment.JoinPersonGroupReducer" />
26
				<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text" />
27
				<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text" />
28

    
29
				<!-- MISC -->
30
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
31
				<PROPERTY key="mapreduce.map.speculative" value="false" />
32
				<PROPERTY key="mapred.reduce.tasks" value="10" />
33

    
34
				<PROPERTY key="hash.values.csv" value="manghip,pmanghi,corog,gcoro" />
35

    
36
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
37

    
38
				<!--  	Uncomment to override the default lib path -->
39
				<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
40

    
41
			</STATIC_CONFIGURATION>
42
			<JOB_INTERFACE>
43
				<PARAM name="mapred.input.dir" required="true" description="input sequence file" />
44
			</JOB_INTERFACE>
45
		</HADOOP_JOB>
46
		<STATUS>
47
			<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
48
			<RUNNING_INSTANCES value="0"/>
49
			<CUMULATIVE_RUN value="0" />
50
		</STATUS>
51
		<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
52
	</BODY>
53
</RESOURCE_PROFILE>
54

    
(68-68/89)