Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="721fd82c-6444-41c9-ba23-5eb0652ddaeb_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="publicationAnalysisJob" type="mapreduce">
11
 			<DESCRIPTION>map only job that analyses publication features</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
        		<!-- MAPPER -->
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.experiment.PublicationAnalysisMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.NullWritable" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.NullWritable" />
22
			
23
				<!-- MISC -->
24
				
25
				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />
26
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />	
27
				<PROPERTY key="mapreduce.map.speculative" value="false" />
28
				<PROPERTY key="mapreduce.reduce.speculative" value="false" />	
29
			
30
				<PROPERTY key="mapred.reduce.tasks" value="0" />
31
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
32
				
33
		<!--  	Uncomment to override the default lib path -->			
34
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
35
        	</STATIC_CONFIGURATION>
36
        	<JOB_INTERFACE>
37
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
38
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
39
        	</JOB_INTERFACE>
40
        	<SCAN>
41
        		<FILTERS operator="MUST_PASS_ALL">
42
        			<FILTER type="prefix" value="50" />
43
        		</FILTERS>
44
        		<FAMILIES>
45
	       			<FAMILY value="result" />
46
        		</FAMILIES>
47
        	</SCAN>
48
        </HADOOP_JOB>
49
        <STATUS>
50
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
51
            <RUNNING_INSTANCES value="0"/>
52
            <CUMULATIVE_RUN value="0" />
53
        </STATUS>
54
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
55
    </BODY>
56
</RESOURCE_PROFILE>
(81-81/85)