Project

General

Profile

1
<RESOURCE_PROFILE>
2
	<HEADER>
3
		<RESOURCE_IDENTIFIER value="baa63c0c-0ff3-4a15-93c1-e361800e9ca8_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="dedupSimilarity2HdfsActionsJob" type="mapreduce">
11
			<DESCRIPTION>map reduce job that scans a given entity type and creates the similarRel graph</DESCRIPTION>
12
			<STATIC_CONFIGURATION>
13

    
14
				<!-- I/O FORMAT -->
15
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
16
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat" />
17

    
18
				<!-- MAPPER -->
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupSimilarityToHdfsActionsMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text" />
22

    
23
				<!-- OUTPUT -->
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.compress.map.output" value="true" />
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.output.compress" value="true"/>
35
				<PROPERTY key="mapred.output.compression.type" value="BLOCK"/>
36

    
37
				<PROPERTY key="mapred.reduce.tasks" value="10" />
38
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
39

    
40
				<!--  	Uncomment to override the default lib path -->
41
				<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
42
			</STATIC_CONFIGURATION>
43
			<JOB_INTERFACE>
44
				<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
45
				<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
46
				<PARAM name="mapred.output.dir" required="true" description="target hbase table" />
47
				<PARAM name="dedup.conf" required="true" description="dedup configuration" />
48
				<PARAM name="rawSetId" required="true" description="raw set identifier" />
49
				<PARAM name="similarityCF" required="true" description="similarity column family name" />
50
			</JOB_INTERFACE>
51
			<SCAN>
52
				<FILTERS operator="MUST_PASS_ALL">
53
					<FILTER type="prefix" param="entityTypeId" />
54
				</FILTERS>
55
				<FAMILIES>
56
					<FAMILY param="entityType" />
57
					<FAMILY value="resultResult_dedupSimilarity_isSimilarTo" />
58
					<FAMILY value="organizationOrganization_dedupSimilarity_isSimilarTo" />
59
				</FAMILIES>
60
			</SCAN>
61
		</HADOOP_JOB>
62
		<STATUS>
63
			<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
64
			<RUNNING_INSTANCES value="0"/>
65
			<CUMULATIVE_RUN value="0" />
66
		</STATUS>
67
		<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
68
	</BODY>
69
</RESOURCE_PROFILE>
(35-35/89)