Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="719b5d2b-4156-4936-bbc3-41d908ec3c57_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="elastisearchTestJob" type="mapreduce">
11
 			<DESCRIPTION>map only job that indexes over ES</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.elasticsearch.hadoop.mr.EsOutputFormat" />
17
        	
18
        		<!-- MAPPER -->
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.es.ElasticsearchFeedMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.NullWritable" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.BytesWritable" />
22
				
23
			
24
				<!-- MISC -->
25
				<PROPERTY key="es.nodes" value="146.48.87.110:9200" />
26
				<PROPERTY key="es.resource" value="openaire/oaf" />
27
				<PROPERTY key="es.input.json" value="yes" />
28
				
29
				<PROPERTY key="mapred.reduce.tasks" value="0" />
30

    
31
				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />	
32
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />	
33
	
34
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
35
				
36
		<!--  	Uncomment to override the default lib path -->			
37
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
38
        	</STATIC_CONFIGURATION>
39
        	<JOB_INTERFACE>
40
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
41
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
42
        		
43
        		<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs" />         		
44
        	</JOB_INTERFACE>
45
        	<SCAN>
46
        		<FILTERS />
47
        		<FAMILIES />
48
        	</SCAN>
49
        </HADOOP_JOB>
50
        <STATUS>
51
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
52
            <RUNNING_INSTANCES value="0"/>
53
            <CUMULATIVE_RUN value="0" />
54
        </STATUS>
55
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
56
    </BODY>
57
</RESOURCE_PROFILE>
(39-39/89)