Revision 52963
Added by Claudio Atzori about 5 years ago
modules/dnet-openaireplus-profiles/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/predatoryJournalsJob.xml | ||
---|---|---|
1 |
<RESOURCE_PROFILE> |
|
2 |
<HEADER> |
|
3 |
<RESOURCE_IDENTIFIER value="bab7a0b8-66b4-4e1a-a8d4-0bb9b3493f90_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="predatoryJournalsJob" type="mapreduce"> |
|
11 |
<DESCRIPTION>map reduce job that counts the number of publications from journals linked to EC projects</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.mapreduce.lib.output.TextOutputFormat" /> |
|
17 |
|
|
18 |
<!-- MAPPER --> |
|
19 |
<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.misc.PredatoryJournalsMapper" /> |
|
20 |
<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" /> |
|
21 |
<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" /> |
|
22 |
|
|
23 |
<!-- REDUCER --> |
|
24 |
<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.misc.PredatoryJournalsReducer" /> |
|
25 |
<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text" /> |
|
26 |
<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text" /> |
|
27 |
|
|
28 |
<!-- MISC --> |
|
29 |
<PROPERTY key="mapred.output.compress" value="false" /> |
|
30 |
<PROPERTY key="mapred.compress.map.output" value="true" /> |
|
31 |
<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" /> |
|
32 |
<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" /> |
|
33 |
<PROPERTY key="mapreduce.map.speculative" value="false" /> |
|
34 |
<PROPERTY key="mapreduce.reduce.speculative" value="false" /> |
|
35 |
|
|
36 |
<PROPERTY key="mapred.reduce.tasks" value="1" /> |
|
37 |
<PROPERTY key="dfs.blocksize" value="256M" /> |
|
38 |
|
|
39 |
<!-- <PROPERTY key="user.name" value="dnet" /> --> |
|
40 |
|
|
41 |
<!-- Uncomment to override the default lib path --> |
|
42 |
<PROPERTY key="job.lib" value="/lib/dnet/snapshots/dnet-mapreduce-jobs-assembly-1.1.4-BETA-SNAPSHOT-predatoryJournals.jar"/> |
|
43 |
</STATIC_CONFIGURATION> |
|
44 |
<JOB_INTERFACE> |
|
45 |
<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" /> |
|
46 |
<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" /> |
|
47 |
<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs" /> |
|
48 |
</JOB_INTERFACE> |
|
49 |
<SCAN> |
|
50 |
<FILTERS /> |
|
51 |
<FAMILIES /> |
|
52 |
</SCAN> |
|
53 |
</HADOOP_JOB> |
|
54 |
<STATUS> |
|
55 |
<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/> |
|
56 |
<RUNNING_INSTANCES value="0"/> |
|
57 |
<CUMULATIVE_RUN value="0" /> |
|
58 |
</STATUS> |
|
59 |
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS> |
|
60 |
</BODY> |
|
61 |
</RESOURCE_PROFILE> |
Also available in: Unified diff
added jobs for predatory journal analysis