Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="7a90d36d-88a1-4c72-8ccb-b8a92a1e6663_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="bulkTaggingJob" type="mapreduce">
11
 			<DESCRIPTION>map reduce job that bulk tags results according to a given configuration</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.hbase.mapreduce.TableOutputFormat" />
17

    
18
        		<!-- MAPPER -->
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.bulktag.BulkTaggingMapper" />
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.io.Writable" />
22

    
23
				<!-- MISC -->
24
				<PROPERTY key="mapred.compress.map.output" value="true" />
25
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
26
				<PROPERTY key="mapreduce.map.speculative" value="false" />
27

    
28
				<PROPERTY key="mapred.reduce.tasks" value="0" />
29
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
30

    
31
				<!-- trust for hte bulktagging operation -->
32
				<PROPERTY key="bulktagging.trust" value="0.85"/>
33

    
34
		<!--  	Uncomment to override the default lib path -->
35
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
36
        	</STATIC_CONFIGURATION>
37
        	<JOB_INTERFACE>
38
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
39
        		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
40
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
41
        		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />
42
        		<PARAM name="tagging.conf" required="true" description="bulk tagging configuration" />
43
				<PARAM name="tagging.enabled" required="true" description="enables/disables writing operations on hbase"/>
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>
(6-6/71)