Project

General

Profile

« Previous | Next » 

Revision 49814

integrated exportSummaryRecordsJob profile from dnet40

View differences:

modules/dnet-openaireplus-profiles/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/exportSummaryRecordsJob.xml
1
<RESOURCE_PROFILE>
2
	<HEADER>
3
		<RESOURCE_IDENTIFIER
4
				value="02db1002-894f-48df-9f50-8977ee9866ca_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="exportSummaryJob" type="mapreduce">
12
			<DESCRIPTION>map only job that exports the infospace as summary xml records, depending on the xslt transformation given as input</DESCRIPTION>
13
			<STATIC_CONFIGURATION>
14

  
15
				<!-- I/O FORMAT -->
16
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat"/>
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.ExportSimplifiedRecordsMapper"/>
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="mapred.input.dir" required="true" description="source hbase table"/>
40
				<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs"/>
41
				<PARAM name="xslt" required="true" description="target sequence file on hdfs"/>
42
			</JOB_INTERFACE>
43
			<SCAN/>
44
		</HADOOP_JOB>
45
		<STATUS>
46
			<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
47
			<RUNNING_INSTANCES value="0"/>
48
			<CUMULATIVE_RUN value="0"/>
49
		</STATUS>
50
		<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
51
	</BODY>
52
</RESOURCE_PROFILE>

Also available in: Unified diff