Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="2af384c6-9118-426d-9394-d7bbc42d707c_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="dedupAnchorStatsJob" type="mapreduce">
11
 			<DESCRIPTION>map only job that calculate statistics on the person anchors</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.NullOutputFormat" />
17

    
18
        	
19
        		<!-- MAPPER -->
20
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.experiment.AnchorStatsMapper" />
21
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
22
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text" />
23
				
24
				<!-- JOB GLOBAL -->		
25
                <PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.NullWritable" />
26
                <PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.NullWritable"/>
27
				
28
				<!-- MISC -->					
29
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />	
30
				<PROPERTY key="mapreduce.map.speculative" value="false" />		
31
				<PROPERTY key="mapred.reduce.tasks" value="0" />
32
				<!--<PROPERTY key="mapred.fairscheduler.pool" value="solr"/> -->
33
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
34
				
35
		<!--  	Uncomment to override the default lib path -->			
36
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
37
        	</STATIC_CONFIGURATION>
38
        	<JOB_INTERFACE>
39
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
40
	       		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
41
        	</JOB_INTERFACE>
42
        	<SCAN>
43
        		<FILTERS operator="MUST_PASS_ALL">
44
        			<FILTER type="prefix" param="entityTypeId" />
45
        		</FILTERS>
46
        		<FAMILIES>
47
        			<FAMILY param="entityType" />
48
        		</FAMILIES>
49
        	</SCAN>
50
        </HADOOP_JOB>
51
        <STATUS>
52
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
53
            <RUNNING_INSTANCES value="0"/>
54
            <CUMULATIVE_RUN value="0" />
55
        </STATUS>
56
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
57
    </BODY>
58
</RESOURCE_PROFILE>
(11-11/55)