Project

General

Profile

1
<RESOURCE_PROFILE>
2
	<HEADER>
3
		<RESOURCE_IDENTIFIER
4
				value="d5a7c415-932b-442e-91c2-46f648945ac2_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
5
		<RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
6
		<RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
7
		<RESOURCE_URI value=""/>
8
		<DATE_OF_CREATION value="2001-12-31T12:00:00"/>
9
	</HEADER>
10
	<BODY>
11
		<HADOOP_JOB name="exportIdentifiersJob" type="mapreduce">
12
			<DESCRIPTION>map only job that exports the publication identifiers as json</DESCRIPTION>
13
			<STATIC_CONFIGURATION>
14

    
15
				<!-- I/O FORMAT -->
16
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat"/>
17
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.TextOutputFormat"/>
18

    
19
				<!-- MAPPER -->
20
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dataexport.ExportResultIdentifiersMapper"/>
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
				<!-- MISC -->
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="dfs.blocksize" value="256M"/>
31

    
32
				<PROPERTY key="mapred.reduce.tasks" value="1"/>
33
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
34

    
35
				<!--  	Uncomment to override the default lib path -->
36
				<PROPERTY key="job.lib" value="/lib/dnet/snapshots/dnet-mapreduce-jobs-assembly-0.0.8.5-SNAPSHOT-exportIds.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

    
42
				<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs"/>
43
			</JOB_INTERFACE>
44
			<SCAN>
45
				<FILTERS operator="MUST_PASS_ALL">
46
					<FILTER type="prefix" value="50"/>
47
				</FILTERS>
48
				<FAMILIES>
49
					<FAMILY value="result"/>
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>
(39-39/75)