1
|
<RESOURCE_PROFILE>
|
2
|
<HEADER>
|
3
|
<RESOURCE_IDENTIFIER value="b4e994d6-0bc3-4010-8419-259603d3ca8d_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
|
4
|
<RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
|
5
|
<RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
|
6
|
<RESOURCE_URI value=""/>
|
7
|
<DATE_OF_CREATION value="2019-07-17T10:26:14+00:00"/>
|
8
|
</HEADER>
|
9
|
<BODY>
|
10
|
<HADOOP_JOB name="informationSpaceMergedUpdatesExportJob" type="mapreduce">
|
11
|
<DESCRIPTION>map only job that exports the whole information space table as json with the updates already merged within the body</DESCRIPTION>
|
12
|
<STATIC_CONFIGURATION><!-- I/O FORMAT -->
|
13
|
<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat"/>
|
14
|
<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat"/><!-- MAPPER -->
|
15
|
<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dataexport.ExportInformationSpaceMergedUpdatesMapper"/>
|
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"/>
|
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.reduce.tasks.speculative.execution" value="false"/>
|
21
|
<PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/>
|
22
|
<PROPERTY key="mapreduce.map.speculative" value="false"/>
|
23
|
<PROPERTY key="mapreduce.reduce.speculative" value="false"/>
|
24
|
<PROPERTY key="dfs.blocksize" value="256M"/>
|
25
|
<PROPERTY key="mapred.compress.map.output" value="true"/>
|
26
|
<PROPERTY key="mapred.output.compress" value="true"/>
|
27
|
<PROPERTY key="mapred.output.compression.type" value="BLOCK"/>
|
28
|
<PROPERTY key="mapred.reduce.tasks" value="0"/><!-- <PROPERTY key="user.name" value="dnet" /> --><!-- Uncomment to override the default lib path --><!-- <PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
|
29
|
</STATIC_CONFIGURATION>
|
30
|
<JOB_INTERFACE>
|
31
|
<PARAM description="source hbase table" name="hbase.mapred.inputtable" required="true"/>
|
32
|
<PARAM description="source hbase table" name="hbase.mapreduce.inputtable" required="true"/>
|
33
|
<PARAM description="target sequence file on hdfs" name="mapred.output.dir" required="true"/>
|
34
|
</JOB_INTERFACE>
|
35
|
<SCAN>
|
36
|
<FILTERS/>
|
37
|
<FAMILIES/>
|
38
|
</SCAN>
|
39
|
</HADOOP_JOB>
|
40
|
<STATUS>
|
41
|
<LAST_SUBMISSION_DATE value="2019-07-02T10:37:48+00:00"/>
|
42
|
<RUNNING_INSTANCES value="0"/>
|
43
|
<CUMULATIVE_RUN value="33"/>
|
44
|
</STATUS>
|
45
|
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
|
46
|
</BODY>
|
47
|
</RESOURCE_PROFILE>
|