Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="675f1436-205a-4b19-8b6b-35e1c17fb125_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="dedupCandidateScanJob" 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.hbase.mapreduce.TableOutputFormat" />				
17
        	
18
        		<!-- MAPPER -->
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
22
				
23
				<!-- REDUCER -->
24
				<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupReducer" />
25
				<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />				
26
				<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Writable" />				
27
				
28
				<!-- MISC -->				
29
				<PROPERTY key="mapred.compress.map.output" value="true" />	
30
				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />	
31
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />	
32
				<PROPERTY key="mapreduce.map.speculative" value="false" />		
33
				<PROPERTY key="mapreduce.reduce.speculative" value="false" />				
34
			
35
				<PROPERTY key="mapred.reduce.tasks" value="100" />
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
        	</STATIC_CONFIGURATION>
41
        	<JOB_INTERFACE>
42
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
43
        		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
44
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
45
        		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />
46
        		<PARAM name="dedup.pace.conf" required="true" description="dedup pace configuration" />
47
        		<PARAM name="dedup.wf.conf" required="true" description="dedup workflow configuration" />        		
48
        	</JOB_INTERFACE>
49
        	<SCAN>
50
        		<FILTERS operator="MUST_PASS_ALL">
51
        			<FILTER type="prefix" param="entityTypeId" />
52
        		</FILTERS>
53
        		<FAMILIES>
54
        			<FAMILY param="entityType" />
55
           			<FAMILY value="resultResult_dedup_merges" />
56
        			<FAMILY value="personPerson_dedup_merges" />
57
     				<FAMILY value="organizationOrganization_dedup_merges" />        			
58
        		</FAMILIES>
59
        	</SCAN>
60
        </HADOOP_JOB>
61
        <STATUS>
62
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
63
            <RUNNING_INSTANCES value="0"/>
64
            <CUMULATIVE_RUN value="0" />
65
        </STATUS>
66
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
67
    </BODY>
68
</RESOURCE_PROFILE>
(2-2/6)