Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="4e16c8dd-8944-4266-8b5c-62e4b26e3090_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="buildMergedToAnchorMapJob" type="mapreduce">
11
 			<DESCRIPTION>map reduce job that builds a map {merged author id --> anchorId}</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.TextOutputFormat" />				
17
        	
18
        		<!-- MAPPER -->
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.gt.BuildMergedAnchorMapMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text" />
22
				
23
				<!-- REDUCER -->
24
				<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.gt.BuildMergedAnchorMapReducer" />
25
				<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text" />				
26
				<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text" />				
27
				
28
				<!-- MISC -->
29
				<PROPERTY key="mapred.output.compress" value="false" />
30
				<PROPERTY key="mapred.compress.map.output" value="true" />	
31
				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />	
32
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />	
33
				<PROPERTY key="mapreduce.map.speculative" value="false" />		
34
				<PROPERTY key="mapreduce.reduce.speculative" value="false" />
35
							
36
				<PROPERTY key="mapred.reduce.tasks" value="1" />
37
				<PROPERTY key="dfs.blocksize" value="256M" />
38
				<PROPERTY key="mapred.textoutputformat.separator" value="=" />
39
				
40
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
41
				
42
		<!--  	Uncomment to override the default lib path -->			
43
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
44
        	</STATIC_CONFIGURATION>
45
        	<JOB_INTERFACE>
46
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
47
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
48
        		<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs" /> 
49
        	</JOB_INTERFACE>
50
        	<SCAN>
51
                <FILTERS operator="MUST_PASS_ALL">
52
                    <FILTER type="prefix" value="30"/>
53
                </FILTERS>
54
                <FAMILIES>
55
                    <FAMILY value="person"/>
56
	                <FAMILY value="personPerson_dedup_isMergedIn" />
57
                </FAMILIES>
58
        	</SCAN>
59
        </HADOOP_JOB>
60
        <STATUS>
61
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
62
            <RUNNING_INSTANCES value="0"/>
63
            <CUMULATIVE_RUN value="0" />
64
        </STATUS>
65
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
66
    </BODY>
67
</RESOURCE_PROFILE>
(5-5/66)