Project

General

Profile

1
<RESOURCE_PROFILE>
2
	<HEADER>
3
		<RESOURCE_IDENTIFIER
4
				value="28e8d911-87e5-4f39-9ce9-0d445126c75f_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="connectedComponentsJob" type="mapreduce">
12
			<DESCRIPTION>map reduce job joins all the vertex ids to build the connected components in the graph</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.hbase.mapreduce.TableOutputFormat"/>
18

    
19
				<!-- MAPPER -->
20
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.cc.ConnectedComponentsMapper"/>
21
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text"/>
22
				<PROPERTY key="mapred.mapoutput.value.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.cc.VertexWritable"/>
23

    
24
				<!-- REDUCER -->
25
				<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.cc.ConnectedComponentsReducer"/>
26
				<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable"/>
27
				<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.hbase.client.Put"/>
28

    
29
				<!-- MISC -->
30
				<PROPERTY key="mapred.compress.map.output" value="false"/>
31
				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false"/>
32
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/>
33
				<PROPERTY key="mapreduce.map.speculative" value="false"/>
34
				<PROPERTY key="mapreduce.reduce.speculative" value="false"/>
35

    
36
				<PROPERTY key="mapred.reduce.tasks" value="100"/>
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="mapred.input.dir" required="true" description="source path on hdfs"/>
44
				<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table"/>
45
			</JOB_INTERFACE>
46
			<SCAN>
47
				<FILTERS/>
48
				<FAMILIES/>
49
			</SCAN>
50
		</HADOOP_JOB>
51
		<STATUS>
52
			<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
53
			<RUNNING_INSTANCES value="0"/>
54
			<CUMULATIVE_RUN value="0"/>
55
		</STATUS>
56
		<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
57
	</BODY>
58
</RESOURCE_PROFILE>
59

    
(12-12/75)