Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="a525f9b5-b3e5-495d-921a-a0abe734039d_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="brokerAdditionJob" type="mapreduce">
11
 			<DESCRIPTION>map job that calculates the addition events</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.TextOutputFormat"/>
17

    
18
		        <!-- MAPPER -->
19
		        <PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.broker.add.AdditionMapper"/>
20
		        <PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text"/>
21
		        <PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text"/>
22

    
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="10"/>
33

    
34
		        <PROPERTY key="broker.datasource.type.whitelist" value="pubsrepository::unknown,pubsrepository::institutional,pubsrepository::thematic"/>
35

    
36
		        <!-- <PROPERTY key="user.name" value="dnet" /> -->
37
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
38

    
39
		        <PROPERTY key="mapred.fairscheduler.pool" value="solr"/>
40

    
41
		<!--  	Uncomment to override the default lib path -->			
42
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
43
        	</STATIC_CONFIGURATION>
44
        	<JOB_INTERFACE>
45
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
46
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
47
        		
48
        		<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs" />         		
49
        	</JOB_INTERFACE>
50
        	<SCAN>
51
        		<FILTERS operator="MUST_PASS_ALL">
52
        			<FILTER type="prefix" value="50" />
53
        		</FILTERS>
54
        		<FAMILIES>
55
	       			<FAMILY value="result" />
56
        			<FAMILY value="resultProject_outcome_isProducedBy" />
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>
(1-1/55)