Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="b7d51a07-6996-4841-9a4a-685a044638e3_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="informationSpaceImportJob" type="mapreduce">
11
 			<DESCRIPTION>map only job that import the whole information space table from a sequence file holding a json dump</DESCRIPTION>
12
        	<STATIC_CONFIGURATION>
13
        	
14
				<!-- I/O FORMAT -->
15
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat" />
16
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat" />
17
				
18
				<!-- MAPPER -->        	
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dataimport.ImportInformationSpaceDumpMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.client.Put" />
22
			
23
				<!-- MISC -->
24
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
25
				<PROPERTY key="mapreduce.map.speculative" value="false" />
26
				<PROPERTY key="mapred.reduce.tasks" value="0" />
27
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
28
				
29
		<!--  	Uncomment to override the default lib path -->			
30
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
31
        	</STATIC_CONFIGURATION>
32
        	<JOB_INTERFACE>
33
        		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
34
        		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />
35
        		
36
        		<PARAM name="mapred.input.dir" required="true" description="source sequence file on hdfs" />         		
37
        	</JOB_INTERFACE>
38
        	<SCAN>
39
        		<FILTERS />
40
        		<FAMILIES />
41
        	</SCAN>
42
        </HADOOP_JOB>
43
        <STATUS>
44
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
45
            <RUNNING_INSTANCES value="0"/>
46
            <CUMULATIVE_RUN value="0" />
47
        </STATUS>
48
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
49
    </BODY>
50
</RESOURCE_PROFILE>
(57-57/85)