Project

General

Profile

1
<RESOURCE_PROFILE>
2
	<HEADER>
3
		<RESOURCE_IDENTIFIER
4
				value="5907741a-d97e-41ea-9dbe-963209aa58d9_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
5
		<RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
6
		<RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
7
		<RESOURCE_URI value=""/>
8
		<DATE_OF_CREATION value="2001-12-31T12:00:00"/>
9
	</HEADER>
10
	<BODY>
11
		<HADOOP_JOB name="dedupSimilarity2GraphJob" type="mapreduce">
12
			<DESCRIPTION>map only job that scans a given entity type and creates the similarRel graph as adjacency lists</DESCRIPTION>
13
			<STATIC_CONFIGURATION>
14

    
15
				<!-- I/O FORMAT -->
16
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat"/>
17
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat"/>
18

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

    
24
				<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text"/>
25
				<PROPERTY key="mapred.output.value.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.cc.VertexWritable"/>
26

    
27
				<!-- MISC -->
28
				<PROPERTY key="mapred.compress.map.output" value="false"/>
29
				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false"/>
30
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/>
31
				<PROPERTY key="mapreduce.map.speculative" value="false"/>
32
				<PROPERTY key="mapreduce.reduce.speculative" value="false"/>
33

    
34
				<PROPERTY key="mapred.reduce.tasks" value="0"/>
35
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
36

    
37
				<!--  	Uncomment to override the default lib path -->
38
				<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
39
			</STATIC_CONFIGURATION>
40
			<JOB_INTERFACE>
41
				<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table"/>
42
				<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table"/>
43
				<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs"/>
44
			</JOB_INTERFACE>
45
			<SCAN>
46
				<FILTERS operator="MUST_PASS_ALL">
47
					<FILTER type="prefix" param="entityTypeId"/>
48
				</FILTERS>
49
				<FAMILIES>
50
					<FAMILY value="resultResult_dedupSimilarity_isSimilarTo"/>
51
					<FAMILY value="organizationOrganization_dedupSimilarity_isSimilarTo"/>
52
				</FAMILIES>
53
			</SCAN>
54
		</HADOOP_JOB>
55
		<STATUS>
56
			<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
57
			<RUNNING_INSTANCES value="0"/>
58
			<CUMULATIVE_RUN value="0"/>
59
		</STATUS>
60
		<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
61
	</BODY>
62
</RESOURCE_PROFILE>
63

    
(35-35/82)