Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="ed8c0a4e-7cf2-49df-bfed-fcfab0699ade_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="prepareIndexDataJob" type="mapreduce">
11
 			<DESCRIPTION>map reduce job that joins the entities on the hbase table and produces a sequence file containig the xml records</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.SequenceFileOutputFormat" />				
17
        	
18
        		<!-- MAPPER -->
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.index.PrepareFeedMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
22
				
23
				<!-- REDUCER -->
24
				<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.index.PrepareFeedReducer" />
25
				<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text" />				
26
				<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text" />				
27
				
28
				<!-- MISC -->
29
				<PROPERTY key="mapred.output.compress" value="false" />						
30
				<PROPERTY key="mapred.compress.map.output" value="true" />	
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="500" />
37
				<PROPERTY key="dfs.blocksize" value="256M" />
38
		        <PROPERTY key="ui.other.datasourcetypes" value="scholarcomminfra, infospace, pubsrepository::mock, entityregistry, entityregistry::projects, entityregistry::repositories, websource"/>
39

    
40
				
41
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
42
				
43
		<!--  	Uncomment to override the default lib path -->			
44
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
45
        	</STATIC_CONFIGURATION>
46
        	<JOB_INTERFACE>
47
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
48
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
49
        		<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs" /> 
50
        		
51
        		<PARAM name="index.entity.links" required="true" description="entity joiner configuration" /> 
52
        		<PARAM name="contextmap" required="true" description="context map (ContextDSResources)" />
53

    
54
	        </JOB_INTERFACE>
55
        	<SCAN>
56
        		<FILTERS />
57
        		<FAMILIES />
58
        	</SCAN>
59
        </HADOOP_JOB>
60
        <STATUS>
61
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
62
            <RUNNING_INSTANCES value="0"/>
63
            <CUMULATIVE_RUN value="0" />
64
        </STATUS>
65
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
66
    </BODY>
67
</RESOURCE_PROFILE>
(49-49/55)