Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="93078a14-40f4-49f0-9206-a16e6d4e342d_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="informationSpaceConfigurableExportJob" type="mapreduce">
11
 			<DESCRIPTION>map only job that exports the whole information space table as json</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.dataexport.ConfigurableExportMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text" />
22
			
23
			
24
				<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text" />				
25
				<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text" />	
26
			
27
				<!-- MISC -->
28

    
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="dfs.blocksize" value="256M" />
35
				
36
				<PROPERTY key="mapred.reduce.tasks" value="1" />
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.mapreduce.inputtable" required="true" description="source hbase table" />
45
        		
46
        		<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs" />         		
47
        	</JOB_INTERFACE>
48
        	<SCAN>
49
        		<FILTERS />
50
        		<FAMILIES />
51
        	</SCAN>
52
        </HADOOP_JOB>
53
        <STATUS>
54
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
55
            <RUNNING_INSTANCES value="0"/>
56
            <CUMULATIVE_RUN value="0" />
57
        </STATUS>
58
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
59
    </BODY>
60
</RESOURCE_PROFILE>
(48-48/71)