Project

General

Profile

1 39222 claudio.at
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="62c1e44c-14b6-4639-9083-29bf432e6759_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="informationSpaceExportJob" type="mapreduce">
11
 			<DESCRIPTION>map only job that exports the whole information space table as json</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.dataexport.ExportInformationSpaceMapper" />
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
				<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text" />
25
				<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text" />
26
27
				<!-- MISC -->
28
29
				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />
30
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
31
				<PROPERTY key="mapreduce.map.speculative" value="false" />
32
				<PROPERTY key="mapreduce.reduce.speculative" value="false" />
33
34
				<PROPERTY key="dfs.blocksize" value="256M" />
35 39268 claudio.at
36 39296 claudio.at
				<PROPERTY key="mapred.compress.map.output" value="true" />
37 39268 claudio.at
				<PROPERTY key="mapred.output.compress" value="true" />
38 39296 claudio.at
				<PROPERTY key="mapred.output.compression.type" value="BLOCK" />
39
40 39222 claudio.at
41
				<PROPERTY key="mapred.reduce.tasks" value="0" />
42
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
43
44
		<!--  	Uncomment to override the default lib path -->
45
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
46
        	</STATIC_CONFIGURATION>
47
        	<JOB_INTERFACE>
48
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
49
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
50
51
        		<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs" />
52
        	</JOB_INTERFACE>
53
        	<SCAN>
54
        		<FILTERS />
55
        		<FAMILIES />
56
        	</SCAN>
57
        </HADOOP_JOB>
58
        <STATUS>
59
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
60
            <RUNNING_INSTANCES value="0"/>
61
            <CUMULATIVE_RUN value="0" />
62
        </STATUS>
63
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
64
    </BODY>
65
</RESOURCE_PROFILE>