1 |
34883
|
claudio.at
|
<RESOURCE_PROFILE>
|
2 |
|
|
<HEADER>
|
3 |
49702
|
claudio.at
|
<RESOURCE_IDENTIFIER value="17dd747e-f5f2-45d5-8554-9f70343bfe55_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
|
4 |
34883
|
claudio.at
|
<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 |
49702
|
claudio.at
|
<HADOOP_JOB name="brokerEnrichmentProjectsJob" type="mapreduce">
|
11 |
43187
|
claudio.at
|
<DESCRIPTION>map reduce job that calculates the enrichment events based on the publications dedup results</DESCRIPTION>
|
12 |
34883
|
claudio.at
|
<STATIC_CONFIGURATION>
|
13 |
|
|
|
14 |
43187
|
claudio.at
|
<!-- 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 |
34883
|
claudio.at
|
|
18 |
43187
|
claudio.at
|
<!-- MAPPER -->
|
19 |
49702
|
claudio.at
|
<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.broker.enrich.ProjectEnrichmentMapper"/>
|
20 |
43187
|
claudio.at
|
<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable"/>
|
21 |
|
|
<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable"/>
|
22 |
|
|
|
23 |
|
|
<!-- REDUCER -->
|
24 |
49702
|
claudio.at
|
<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.broker.enrich.ProjectEnrichmentReducer" />
|
25 |
43187
|
claudio.at
|
<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.reduce.tasks.speculative.execution" value="false"/>
|
30 |
|
|
<PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/>
|
31 |
|
|
<PROPERTY key="mapreduce.map.speculative" value="false"/>
|
32 |
|
|
<PROPERTY key="mapreduce.reduce.speculative" value="false"/>
|
33 |
|
|
|
34 |
|
|
<PROPERTY key="dfs.blocksize" value="256M"/>
|
35 |
|
|
|
36 |
|
|
<PROPERTY key="mapred.reduce.tasks" value="1"/>
|
37 |
|
|
|
38 |
|
|
<PROPERTY key="broker.datasource.id.whitelist" value=""/>
|
39 |
|
|
<PROPERTY key="broker.datasource.id.blacklist" value=""/>
|
40 |
43379
|
claudio.at
|
<PROPERTY key="broker.datasource.untrusted.oa.list" value="opendoar____::8b6dd7db9af49e67306feb59a8bdc52c"/>
|
41 |
43187
|
claudio.at
|
|
42 |
|
|
<PROPERTY key="broker.datasource.type.whitelist" value="pubsrepository::unknown,pubsrepository::institutional,pubsrepository::thematic"/>
|
43 |
|
|
|
44 |
|
|
<!-- <PROPERTY key="user.name" value="dnet" /> -->
|
45 |
34883
|
claudio.at
|
<!-- <PROPERTY key="user.name" value="dnet" /> -->
|
46 |
|
|
|
47 |
|
|
<!-- Uncomment to override the default lib path -->
|
48 |
|
|
<!-- <PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
|
49 |
|
|
</STATIC_CONFIGURATION>
|
50 |
|
|
<JOB_INTERFACE>
|
51 |
|
|
<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
|
52 |
|
|
<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
|
53 |
|
|
|
54 |
|
|
<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs" />
|
55 |
|
|
</JOB_INTERFACE>
|
56 |
|
|
<SCAN>
|
57 |
|
|
<FILTERS operator="MUST_PASS_ALL">
|
58 |
49702
|
claudio.at
|
<FILTER type="prefix" value="40" />
|
59 |
43187
|
claudio.at
|
<FILTER type="prefix" value="50" />
|
60 |
34883
|
claudio.at
|
</FILTERS>
|
61 |
|
|
<FAMILIES>
|
62 |
43187
|
claudio.at
|
<FAMILY value="result" />
|
63 |
49702
|
claudio.at
|
<FAMILY value="project" />
|
64 |
43187
|
claudio.at
|
<FAMILY value="resultResult_dedup_isMergedIn" />
|
65 |
49702
|
claudio.at
|
<FAMILY value="resultProject_outcome_produces" />
|
66 |
|
|
<FAMILY value="resultProject_outcome_isProducedBy" />
|
67 |
34883
|
claudio.at
|
</FAMILIES>
|
68 |
|
|
</SCAN>
|
69 |
|
|
</HADOOP_JOB>
|
70 |
|
|
<STATUS>
|
71 |
|
|
<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
|
72 |
|
|
<RUNNING_INSTANCES value="0"/>
|
73 |
|
|
<CUMULATIVE_RUN value="0" />
|
74 |
|
|
</STATUS>
|
75 |
|
|
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
|
76 |
|
|
</BODY>
|
77 |
|
|
</RESOURCE_PROFILE>
|