Project

General

Profile

1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="5b99d940-7477-40db-a458-549d301419c5_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="dedupFixRelationsJob" type="mapreduce">
11
            <DESCRIPTION>map reduce job that redirects the rels to fix the graph consistency</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.dedup.fixrelation.DedupFixRelationMapper"/>
20
                <PROPERTY key="mapreduce.mapoutput.key.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.fixrelation.Key"/>
21
                <PROPERTY key="mapred.mapoutput.key.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.fixrelation.Key"/>
22
                <PROPERTY key="mapreduce.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable"/>
23
                <PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable"/>
24

    
25
                <!-- PARTITIONER / SECONDARY SORTING -->
26
                <PROPERTY key="mapreduce.partitioner.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.fixrelation.KeyPartitioner"/>
27
                <PROPERTY key="mapred.partitioner.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.fixrelation.KeyPartitioner"/>
28
                <PROPERTY key="mapred.output.value.groupfn.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.fixrelation.KeyGroupingComparator"/>
29
                <PROPERTY key="mapreduce.output.value.groupfn.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.fixrelation.KeyGroupingComparator"/>
30

    
31
                <!-- REDUCER -->
32
                <PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.fixrelation.DedupFixRelationReducer"/>
33
                <PROPERTY key="mapreduce.output.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable"/>
34
                <PROPERTY key="mapreduce.output.value.class" value="org.apache.hadoop.io.Writable"/>
35

    
36
                <!-- MISC -->
37
                <PROPERTY key="mapreduce.reduce.tasks.speculative.execution" value="false"/>
38
                <PROPERTY key="mapreduce.map.tasks.speculative.execution" value="false"/>
39
                <PROPERTY key="mapreduce.map.speculative" value="false"/>
40
                <PROPERTY key="mapreduce.reduce.speculative" value="false"/>
41
                <PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false"/>
42
                <PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/>
43
                <PROPERTY key="mapred.map.speculative" value="false"/>
44
                <PROPERTY key="mapred.reduce.speculative" value="false"/>
45
                <PROPERTY key="mapred.reduce.tasks" value="100"/>
46
            </STATIC_CONFIGURATION>
47
            <JOB_INTERFACE>
48
                <PARAM description="source hbase table" name="hbase.mapreduce.inputtable" required="true"/>
49
                <PARAM description="target hbase table" name="hbase.mapreduce.outputtable" required="true"/>
50
                <PARAM description="ontology values fetched from the information system" name="ontologies" required="true"/>
51
                <PARAM description="declares the sub entity type to work on" name="subentitytype" required="true"/>
52
            </JOB_INTERFACE>
53
            <SCAN caching="10">
54
                <FILTERS/>
55
                <FAMILIES/>
56
            </SCAN>
57
        </HADOOP_JOB>
58
        <STATUS>
59
            <LAST_SUBMISSION_DATE value="2018-11-27T12:56:57+01:00"/>
60
            <RUNNING_INSTANCES value="0"/>
61
            <CUMULATIVE_RUN value="44"/>
62
        </STATUS>
63
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
64
    </BODY>
65
</RESOURCE_PROFILE>
(23-23/75)