Revision 53780
Added by Claudio Atzori almost 5 years ago
modules/dnet-openaireplus-profiles/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/importCrossrefJob.xml | ||
---|---|---|
1 |
<RESOURCE_PROFILE> |
|
2 |
<HEADER> |
|
3 |
<RESOURCE_IDENTIFIER |
|
4 |
value="f96a082f-c10b-40e2-beb3-a1b1ffdb4a08_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/> |
|
5 |
<RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/> |
|
6 |
<RESOURCE_KIND value="HadoopJobConfigurationDSResources"/> |
|
7 |
<RESOURCE_URI value=""/> |
|
8 |
<DATE_OF_CREATION value="2001-12-31T12:00:00"/> |
|
9 |
</HEADER> |
|
10 |
<BODY> |
|
11 |
<HADOOP_JOB name="importCrossrefJob" type="mapreduce"> |
|
12 |
<DESCRIPTION>map reduce job that import the scholexplorer dump into actions</DESCRIPTION> |
|
13 |
<STATIC_CONFIGURATION> |
|
14 |
|
|
15 |
<!-- I/O FORMAT --> |
|
16 |
<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.TextInputFormat"/> |
|
17 |
<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat"/> |
|
18 |
|
|
19 |
<!-- MAPPER --> |
|
20 |
<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dataimport.CrossRefImportMapper"/> |
|
21 |
<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text"/> |
|
22 |
<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text"/> |
|
23 |
|
|
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.compress.map.output" value="false"/> |
|
29 |
<PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/> |
|
30 |
<PROPERTY key="mapreduce.map.speculative" value="false"/> |
|
31 |
|
|
32 |
<PROPERTY key="mapred.reduce.tasks" value="0"/> |
|
33 |
<PROPERTY key="dfs.blocksize" value="256M"/> |
|
34 |
|
|
35 |
<!-- Crossref Mapper Properties --> |
|
36 |
<PROPERTY key="setName" value="crossref-dump"/> |
|
37 |
<PROPERTY key="agentId" value="dnet"/> |
|
38 |
<PROPERTY key="agentName" value="D-Net"/> |
|
39 |
<PROPERTY key="invisible" value="true"/> |
|
40 |
|
|
41 |
<!-- <PROPERTY key="user.name" value="dnet" /> --> |
|
42 |
|
|
43 |
<!-- Uncomment to override the default lib path --> |
|
44 |
<!-- <PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> --> |
|
45 |
</STATIC_CONFIGURATION> |
|
46 |
<JOB_INTERFACE> |
|
47 |
<PARAM name="mapred.input.dir" required="true" description="source sequence file on hdfs"/> |
|
48 |
<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs"/> |
|
49 |
</JOB_INTERFACE> |
|
50 |
<SCAN> |
|
51 |
<FILTERS/> |
|
52 |
<FAMILIES/> |
|
53 |
</SCAN> |
|
54 |
</HADOOP_JOB> |
|
55 |
<STATUS> |
|
56 |
<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/> |
|
57 |
<RUNNING_INSTANCES value="0"/> |
|
58 |
<CUMULATIVE_RUN value="0"/> |
|
59 |
</STATUS> |
|
60 |
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS> |
|
61 |
</BODY> |
|
62 |
</RESOURCE_PROFILE> |
|
63 |
|
modules/dnet-openaireplus-profiles/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/importDOIBoostJob.xml | ||
---|---|---|
1 |
<RESOURCE_PROFILE> |
|
2 |
<HEADER> |
|
3 |
<RESOURCE_IDENTIFIER |
|
4 |
value="f96a082f-c10b-40e2-beb3-a1b1ffdb4a08_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/> |
|
5 |
<RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/> |
|
6 |
<RESOURCE_KIND value="HadoopJobConfigurationDSResources"/> |
|
7 |
<RESOURCE_URI value=""/> |
|
8 |
<DATE_OF_CREATION value="2001-12-31T12:00:00"/> |
|
9 |
</HEADER> |
|
10 |
<BODY> |
|
11 |
<HADOOP_JOB name="importDOIBoostJob" type="mapreduce"> |
|
12 |
<DESCRIPTION>map reduce job that import the scholexplorer dump into actions</DESCRIPTION> |
|
13 |
<STATIC_CONFIGURATION> |
|
14 |
|
|
15 |
<!-- I/O FORMAT --> |
|
16 |
<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.TextInputFormat"/> |
|
17 |
<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat"/> |
|
18 |
|
|
19 |
<!-- MAPPER --> |
|
20 |
<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dataimport.DOIBoostImportMapper"/> |
|
21 |
<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text"/> |
|
22 |
<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text"/> |
|
23 |
|
|
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 |
<!-- REDUCER --> |
|
28 |
<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.dataimport.DOIBoostImportReducer" /> |
|
29 |
<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text" /> |
|
30 |
<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text" /> |
|
31 |
|
|
32 |
<!-- MISC --> |
|
33 |
<PROPERTY key="mapred.compress.map.output" value="false"/> |
|
34 |
<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false"/> |
|
35 |
<PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/> |
|
36 |
<PROPERTY key="mapreduce.map.speculative" value="false"/> |
|
37 |
<PROPERTY key="mapreduce.reduce.speculative" value="false"/> |
|
38 |
|
|
39 |
<PROPERTY key="mapred.reduce.tasks" value="16"/> |
|
40 |
<PROPERTY key="dfs.blocksize" value="256M"/> |
|
41 |
|
|
42 |
<PROPERTY key="mapred.output.compress" value="true"/> |
|
43 |
<PROPERTY key="mapred.output.compression.type" value="BLOCK"/> |
|
44 |
|
|
45 |
<!-- Crossref Mapper Properties --> |
|
46 |
<PROPERTY key="setName" value="crossref-dump"/> |
|
47 |
<PROPERTY key="agentId" value="dnet"/> |
|
48 |
<PROPERTY key="agentName" value="D-Net"/> |
|
49 |
<PROPERTY key="invisible" value="true"/> |
|
50 |
|
|
51 |
<!-- <PROPERTY key="user.name" value="dnet" /> --> |
|
52 |
|
|
53 |
<!-- Uncomment to override the default lib path --> |
|
54 |
<!-- <PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> --> |
|
55 |
</STATIC_CONFIGURATION> |
|
56 |
<JOB_INTERFACE> |
|
57 |
<PARAM name="mapred.input.dir" required="true" description="source sequence file on hdfs"/> |
|
58 |
<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs"/> |
|
59 |
<PARAM name="onlyOrganization" required="false" description="boolean value that declares import mode: affiliations or ALL"/> |
|
60 |
</JOB_INTERFACE> |
|
61 |
<SCAN> |
|
62 |
<FILTERS/> |
|
63 |
<FAMILIES/> |
|
64 |
</SCAN> |
|
65 |
</HADOOP_JOB> |
|
66 |
<STATUS> |
|
67 |
<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/> |
|
68 |
<RUNNING_INSTANCES value="0"/> |
|
69 |
<CUMULATIVE_RUN value="0"/> |
|
70 |
</STATUS> |
|
71 |
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS> |
|
72 |
</BODY> |
|
73 |
</RESOURCE_PROFILE> |
|
74 |
|
Also available in: Unified diff
avoid to import non necessary affiliations from DOIBoost