1
|
<RESOURCE_PROFILE>
|
2
|
<HEADER>
|
3
|
<RESOURCE_IDENTIFIER
|
4
|
value="667fe203-ee51-4dff-8c9c-b90e66e96eb4_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
|
5
|
<RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
|
6
|
<RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
|
7
|
<RESOURCE_URI value=""/>
|
8
|
<DATE_OF_CREATION value="2001-12-31T12:00:00"/>
|
9
|
</HEADER>
|
10
|
<BODY>
|
11
|
<HADOOP_JOB name="dedupMarkDeletedEntityJob" type="mapreduce">
|
12
|
<DESCRIPTION>map only job that closes the similarity mesh</DESCRIPTION>
|
13
|
<STATIC_CONFIGURATION>
|
14
|
|
15
|
<!-- I/O FORMAT -->
|
16
|
<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat"/>
|
17
|
<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat"/>
|
18
|
|
19
|
<!-- MAPPER -->
|
20
|
<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupMarkDeletedEntityMapper"/>
|
21
|
<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable"/>
|
22
|
<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.client.Put"/>
|
23
|
|
24
|
<!-- MISC -->
|
25
|
<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false"/>
|
26
|
<PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/>
|
27
|
<PROPERTY key="mapreduce.map.speculative" value="false"/>
|
28
|
<PROPERTY key="mapreduce.reduce.speculative" value="false"/>
|
29
|
|
30
|
<PROPERTY key="mapred.reduce.tasks" value="0"/>
|
31
|
<!-- <PROPERTY key="user.name" value="dnet" /> -->
|
32
|
|
33
|
<!-- Uncomment to override the default lib path -->
|
34
|
<!-- <PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
|
35
|
</STATIC_CONFIGURATION>
|
36
|
<JOB_INTERFACE>
|
37
|
<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table"/>
|
38
|
<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table"/>
|
39
|
<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table"/>
|
40
|
<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table"/>
|
41
|
</JOB_INTERFACE>
|
42
|
<SCAN>
|
43
|
<FILTERS operator="MUST_PASS_ALL">
|
44
|
<FILTER type="prefix" param="entityTypeId"/>
|
45
|
</FILTERS>
|
46
|
<FAMILIES>
|
47
|
<FAMILY param="entityType"/>
|
48
|
<FAMILY value="resultResult_dedup_isMergedIn"/>
|
49
|
<FAMILY value="organizationOrganization_dedup_isMergedIn"/>
|
50
|
</FAMILIES>
|
51
|
</SCAN>
|
52
|
</HADOOP_JOB>
|
53
|
<STATUS>
|
54
|
<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
|
55
|
<RUNNING_INSTANCES value="0"/>
|
56
|
<CUMULATIVE_RUN value="0"/>
|
57
|
</STATUS>
|
58
|
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
|
59
|
</BODY>
|
60
|
</RESOURCE_PROFILE>
|