Project

General

Profile

« Previous | Next » 

Revision 52914

added WF for tranform scholexplorer links into actionsets

View differences:

modules/dnet-openaireplus-profiles/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/importScholexplorerJob.xml
1
<RESOURCE_PROFILE>
2
	<HEADER>
3
		<RESOURCE_IDENTIFIER
4
				value="c26983ce-c18c-42c0-92d7-6fa7c77c4725_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="importScholexplorerJob" type="mapreduce">
12
			<DESCRIPTION>map reduce job that import the scholexplorer dump into actions</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.SequenceFileOutputFormat"/>
18

  
19
				<!-- MAPPER -->
20
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.cc.MindistSearchMapper"/>
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.compress.map.output" value="false"/>
26
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/>
27
				<PROPERTY key="mapreduce.map.speculative" value="false"/>
28

  
29
				<PROPERTY key="mapred.reduce.tasks" value="0"/>
30
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
31

  
32
				<!--  	Uncomment to override the default lib path -->
33
				<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
34
			</STATIC_CONFIGURATION>
35
			<JOB_INTERFACE>
36
				<PARAM name="mapred.input.dir" required="true" description="source sequence file on hdfs"/>
37
				<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs"/>
38
			</JOB_INTERFACE>
39
			<SCAN>
40
				<FILTERS/>
41
				<FAMILIES/>
42
			</SCAN>
43
		</HADOOP_JOB>
44
		<STATUS>
45
			<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
46
			<RUNNING_INSTANCES value="0"/>
47
			<CUMULATIVE_RUN value="0"/>
48
		</STATUS>
49
		<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
50
	</BODY>
51
</RESOURCE_PROFILE>
52

  

Also available in: Unified diff