1
|
<RESOURCE_PROFILE>
|
2
|
<HEADER>
|
3
|
<RESOURCE_IDENTIFIER value="ef17e0f5-a80b-4856-8fc3-6c5b99e52f83_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="filterIndexRecordsJob" type="mapreduce">
|
11
|
<DESCRIPTION>map reduce job that filters the index xml records according to the criteria specified in this profile</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.mapreduce.lib.output.SequenceFileOutputFormat" />
|
17
|
|
18
|
<!-- MAPPER -->
|
19
|
<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dataexport.ExportFilteredResultMapper" />
|
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
|
<!-- JOB GLOBAL -->
|
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
|
<PROPERTY key="mapred.task.timeout" value="1800000"/>
|
29
|
<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
|
30
|
<PROPERTY key="mapreduce.map.speculative" value="false" />
|
31
|
<PROPERTY key="mapred.reduce.tasks" value="16" />
|
32
|
|
33
|
<!-- <PROPERTY key="user.name" value="dnet" /> -->
|
34
|
|
35
|
<!-- Uncomment to override the default lib path -->
|
36
|
<!-- <PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
|
37
|
</STATIC_CONFIGURATION>
|
38
|
<JOB_INTERFACE>
|
39
|
<PARAM name="mapred.input.dir" required="true" description="source sequence file on hdfs" />
|
40
|
<PARAM name="mapred.output.dir" required="true" description="destination path on hdfs for the filtered records" />
|
41
|
|
42
|
<PARAM name="filter.defaultcriteria" required="true" description="default criteria" />
|
43
|
<PARAM name="filter.yearxpath" required="true" description="xpath pointing to the dateofacceptance" />
|
44
|
<PARAM name="filter.fromyear" required="true" description="min year" />
|
45
|
<PARAM name="filter.toyear" required="true" description="max year" />
|
46
|
</JOB_INTERFACE>
|
47
|
<SCAN>
|
48
|
<FILTERS operator="MUST_PASS_ALL">
|
49
|
<FILTER type="prefix" value="50"/>
|
50
|
</FILTERS>
|
51
|
<FAMILIES>
|
52
|
<FAMILY value="result"/>
|
53
|
</FAMILIES>
|
54
|
</SCAN>
|
55
|
</HADOOP_JOB>
|
56
|
<STATUS>
|
57
|
<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
|
58
|
<RUNNING_INSTANCES value="0"/>
|
59
|
<CUMULATIVE_RUN value="0" />
|
60
|
</STATUS>
|
61
|
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
|
62
|
</BODY>
|
63
|
</RESOURCE_PROFILE>
|