1
|
<RESOURCE_PROFILE>
|
2
|
<HEADER>
|
3
|
<RESOURCE_IDENTIFIER value="7edd375c-1d1e-4f2b-81c9-1b7d4f65c233_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
|
4
|
<RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
|
5
|
<RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
|
6
|
<RESOURCE_URI value=""/>
|
7
|
<DATE_OF_CREATION value="2018-10-22T10:34:48+00:00"/>
|
8
|
</HEADER>
|
9
|
<BODY>
|
10
|
<HADOOP_JOB name="invalidRecordDoiExporterJob" type="mapreduce">
|
11
|
<DESCRIPTION>map only job that ensures xml records are valid</DESCRIPTION>
|
12
|
<STATIC_CONFIGURATION><!-- I/O FORMAT -->
|
13
|
<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat"/>
|
14
|
<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.TextOutputFormat"/><!-- MAPPER -->
|
15
|
<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dataimport.GetInvalidXmlRecordsMapper"/>
|
16
|
<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text"/>
|
17
|
<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text"/><!-- JOB GLOBAL -->
|
18
|
<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text"/>
|
19
|
<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text"/><!-- MISC -->
|
20
|
<PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/>
|
21
|
<PROPERTY key="mapreduce.map.speculative" value="false"/>
|
22
|
<PROPERTY key="mapred.reduce.tasks" value="1"/><!-- <PROPERTY key="user.name" value="dnet" /> --><!-- Uncomment to override the default lib path -->
|
23
|
<!-- <PROPERTY key="job.lib" value="/lib/dnet/snapshots/dnet-mapreduce-jobs-assembly-1.1.4-BETA-SNAPSHOT-FilterXmlRecordsMapper.jar"/> -->
|
24
|
<PROPERTY key="mapred.output.compress" value="false"/>
|
25
|
</STATIC_CONFIGURATION>
|
26
|
<JOB_INTERFACE>
|
27
|
<PARAM description="input sequence file" name="mapred.input.dir" required="true"/>
|
28
|
<PARAM description="target file" name="mapred.output.dir" required="true"/>
|
29
|
</JOB_INTERFACE>
|
30
|
</HADOOP_JOB>
|
31
|
<STATUS>
|
32
|
<LAST_SUBMISSION_DATE value="2018-10-22T10:35:08+00:00"/>
|
33
|
<RUNNING_INSTANCES value="0"/>
|
34
|
<CUMULATIVE_RUN value="20"/>
|
35
|
</STATUS>
|
36
|
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
|
37
|
</BODY>
|
38
|
</RESOURCE_PROFILE>
|