Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="ba309300-76f2-40d1-afe3-b77016f443e9_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="exportPersonFullnamesJob" type="mapreduce">
11
 			<DESCRIPTION>map reduce job that exports the person fullnames on a text file on HDFS</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.preprocess.ExportFullnameMapper" />
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.preprocess.ExportFullnameReducer" />
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.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="1" />
35
				<PROPERTY key="dfs.blocksize" value="256M" />
36
				<PROPERTY key="mapred.textoutputformat.separator" value="" />
37
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
38
	
39
		<!--  	Uncomment to override the default lib path -->			
40
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
41
        	</STATIC_CONFIGURATION>
42
        	<JOB_INTERFACE>
43
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
44
        		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
45
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
46
        		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />     
47
        	</JOB_INTERFACE>
48
        	<SCAN caching="10">
49
        		<FILTERS operator="MUST_PASS_ALL">
50
        			<FILTER type="prefix" param="entityTypeId" />
51
        		</FILTERS>
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>
(19-19/85)