Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="6d91b311-a7fd-48ff-98d2-1fed70850e3a_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="coauthorUpdateJob" type="mapreduce">
11
 			<DESCRIPTION>update coauthors using 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.inputformat.class"	value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
17
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat" />			
18
        	
19
        		<!-- MAPPER -->
20
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.gt.CoAuthorUpdateMapper" />
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.client.Put" />
23
				
24
				
25
				<!-- MISC -->
26
				<PROPERTY key="mapred.output.compress" value="false" />
27
				<PROPERTY key="mapred.compress.map.output" value="true" />	
28
				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />	
29
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />	
30
				<PROPERTY key="mapreduce.map.speculative" value="false" />		
31
				<PROPERTY key="mapreduce.reduce.speculative" value="false" />
32
							
33
				<PROPERTY key="mapred.reduce.tasks" value="0" />
34
				
35
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
36
				
37
		<!--  	Uncomment to override the default lib path -->			
38
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
39
        	</STATIC_CONFIGURATION>
40
        	<JOB_INTERFACE>
41
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
42
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
43
        		<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs" /> 
44
        	</JOB_INTERFACE>
45
        	<SCAN>
46
                <FILTERS operator="MUST_PASS_ALL">
47
                    <FILTER type="prefix" value="30"/>
48
                </FILTERS>
49
                <FAMILIES>
50
                    <FAMILY value="person"/>
51
                </FAMILIES>
52
        	</SCAN>
53
        </HADOOP_JOB>
54
        <STATUS>
55
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
56
            <RUNNING_INSTANCES value="0"/>
57
            <CUMULATIVE_RUN value="0" />
58
        </STATUS>
59
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
60
    </BODY>
61
</RESOURCE_PROFILE>
(11-11/75)