Revision 53707
Added by Claudio Atzori over 5 years ago
modules/dnet-openaireplus-profiles/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/xmlRecordCounterJob.xml | ||
---|---|---|
1 |
<RESOURCE_PROFILE> |
|
2 |
<HEADER> |
|
3 |
<RESOURCE_IDENTIFIER value="16845094-5871-4584-b934-590b5d005836_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="xmlRecordCounterJob" type="mapreduce"> |
|
11 |
<DESCRIPTION>map only job that count information from the index xml records</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.NullOutputFormat"/><!-- MAPPER --> |
|
15 |
<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.index.CountXmlRecordsMapper"/> |
|
16 |
<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.NullWritable"/> |
|
17 |
<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.NullWritable"/><!-- JOB GLOBAL --> |
|
18 |
<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.NullWritable"/> |
|
19 |
<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.NullWritable"/><!-- 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="0"/><!-- <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 |
</STATIC_CONFIGURATION> |
|
25 |
<JOB_INTERFACE> |
|
26 |
<PARAM description="input sequence file" name="mapred.input.dir" required="true"/> |
|
27 |
</JOB_INTERFACE> |
|
28 |
</HADOOP_JOB> |
|
29 |
<STATUS> |
|
30 |
<LAST_SUBMISSION_DATE value="2018-10-22T10:35:08+00:00"/> |
|
31 |
<RUNNING_INSTANCES value="0"/> |
|
32 |
<CUMULATIVE_RUN value="20"/> |
|
33 |
</STATUS> |
|
34 |
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS> |
|
35 |
</BODY> |
|
36 |
</RESOURCE_PROFILE> |
modules/dnet-openaireplus-profiles/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/countInfospaceJob.xml | ||
---|---|---|
1 |
<RESOURCE_PROFILE> |
|
2 |
<HEADER> |
|
3 |
<RESOURCE_IDENTIFIER value="1e039496-05ef-4913-a0b2-c6fb93e5df39_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="countInfospaceJob" type="mapreduce"> |
|
11 |
<DESCRIPTION>map reduce job that joins the entities on the hbase table and counts the resulting entities and relationships</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.NullOutputFormat"/><!-- MAPPER --> |
|
15 |
<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.index.PrepareFeedMapper"/> |
|
16 |
<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text"/> |
|
17 |
<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable"/><!-- REDUCER --> |
|
18 |
<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.index.InfospaceCountsReducer"/> |
|
19 |
<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.NullWritable"/> |
|
20 |
<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.NullWritable"/><!-- MISC --> |
|
21 |
<PROPERTY key="mapred.output.compress" value="false"/> |
|
22 |
<PROPERTY key="mapred.compress.map.output" value="true"/> |
|
23 |
<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false"/> |
|
24 |
<PROPERTY key="mapred.map.tasks.speculative.execution" value="false"/> |
|
25 |
<PROPERTY key="mapreduce.map.speculative" value="false"/> |
|
26 |
<PROPERTY key="mapreduce.reduce.speculative" value="false"/> |
|
27 |
<PROPERTY key="mapred.reduce.tasks" value="500"/><!-- <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"/> --> |
|
28 |
</STATIC_CONFIGURATION> |
|
29 |
<JOB_INTERFACE> |
|
30 |
<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table"/> |
|
31 |
<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table"/> |
|
32 |
<PARAM name="index.entity.links" required="true" description="entity joiner configuration"/> |
|
33 |
<PARAM name="contextmap" required="true" description="context map (ContextDSResources)"/> |
|
34 |
</JOB_INTERFACE> |
|
35 |
<SCAN> |
|
36 |
<FILTERS/> |
|
37 |
<FAMILIES/> |
|
38 |
</SCAN> |
|
39 |
</HADOOP_JOB> |
|
40 |
<STATUS> |
|
41 |
<LAST_SUBMISSION_DATE value="2018-11-09T14:21:05+00:00"/> |
|
42 |
<RUNNING_INSTANCES value="0"/> |
|
43 |
<CUMULATIVE_RUN value="4"/> |
|
44 |
</STATUS> |
|
45 |
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS> |
|
46 |
</BODY> |
|
47 |
</RESOURCE_PROFILE> |
Also available in: Unified diff
added job profiles for infoSpace counts