Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="027554bd-3d5c-4c50-9170-90d8c4402bc3_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="dedupRootsToCSVJob" type="mapreduce">
11
 			<DESCRIPTION>map only job that exports the representatives as CSV files</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.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.DedupRootsToCsvMapper" />
21
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
22
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
23
			
24
				<!-- REDUCER -->
25
				<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupRootsToCsvReducer" />
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
				<!-- MUTIPLE OUTPUT -->
30
				<PROPERTY key="mapreduce.multipleoutputs" value="NativeGroups Groups NativeEntities" />
31

    
32
				<PROPERTY key="mapreduce.multipleoutputs.namedOutput.NativeGroups.format" value="org.apache.hadoop.mapreduce.lib.output.TextOutputFormat" />
33
				<PROPERTY key="mapreduce.multipleoutputs.namedOutput.NativeGroups.key" value="org.apache.hadoop.io.Text" />
34
				<PROPERTY key="mapreduce.multipleoutputs.namedOutput.NativeGroups.value" value="org.apache.hadoop.io.Text" />
35
				
36
				<PROPERTY key="mapreduce.multipleoutputs.namedOutput.Groups.format" value="org.apache.hadoop.mapreduce.lib.output.TextOutputFormat" />
37
				<PROPERTY key="mapreduce.multipleoutputs.namedOutput.Groups.key" value="org.apache.hadoop.io.Text" />
38
				<PROPERTY key="mapreduce.multipleoutputs.namedOutput.Groups.value" value="org.apache.hadoop.io.Text" />				
39
				
40
				<PROPERTY key="mapreduce.multipleoutputs.namedOutput.NativeEntities.format" value="org.apache.hadoop.mapreduce.lib.output.TextOutputFormat" />
41
				<PROPERTY key="mapreduce.multipleoutputs.namedOutput.NativeEntities.key" value="org.apache.hadoop.io.Text" />
42
				<PROPERTY key="mapreduce.multipleoutputs.namedOutput.NativeEntities.value" value="org.apache.hadoop.io.Text" />				
43
				
44
				<!-- MISC -->
45
				
46
                <PROPERTY key="mapred.textoutputformat.wrapper" value="#"/>
47
                <PROPERTY key="mapred.textoutputformat.separator" value="!"/>
48

    
49
				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />	
50
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />	
51
				<PROPERTY key="mapreduce.map.speculative" value="false" />
52
				<PROPERTY key="mapreduce.reduce.speculative" value="false" />	
53
			
54
				<PROPERTY key="mapred.reduce.tasks" value="3" />
55
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
56
				
57
		<!--  	Uncomment to override the default lib path -->			
58
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
59
        	</STATIC_CONFIGURATION>
60
        	<JOB_INTERFACE>
61
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
62
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
63
        		
64
        		<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs" />         		
65
        	</JOB_INTERFACE>
66
        	<SCAN>
67
        		<FILTERS operator="MUST_PASS_ALL">
68
        			<FILTER type="prefix" param="entityTypeId" />
69
        		</FILTERS>
70
        		<FAMILIES>
71
	       			<FAMILY param="entityType" />
72
        			<FAMILY value="resultResult_dedup_merges" />
73
        			<FAMILY value="personPerson_dedup_merges" />
74
     				<FAMILY value="organizationOrganization_dedup_merges" />
75
        		</FAMILIES>
76
        	</SCAN>
77
        </HADOOP_JOB>
78
        <STATUS>
79
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
80
            <RUNNING_INSTANCES value="0"/>
81
            <CUMULATIVE_RUN value="0" />
82
        </STATUS>
83
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
84
    </BODY>
85
</RESOURCE_PROFILE>
(28-28/55)