Revision 28161
Added by Eri Katsari over 10 years ago
| modules/dnet-openaire-stats/trunk/src/main/resources/eu/dnetlib/data/mapreduce/hbase/statsExport/statsJob.xml | ||
|---|---|---|
| 1 | 1 |
<RESOURCE_PROFILE> |
| 2 |
<HEADER> |
|
| 3 |
<RESOURCE_IDENTIFIER value="ed8c0a4e-7cf2-49df-bfed-fcfab0699ade_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="StatsJob" type="mapreduce"> |
|
| 11 |
<DESCRIPTION>map reduce job that exports hbase data and prepares them for import to the relation database used for statistics generation</DESCRIPTION> |
|
| 12 |
<STATIC_CONFIGURATION> |
|
| 13 |
|
|
| 14 |
|
|
| 2 |
<HEADER> |
|
| 3 |
<RESOURCE_IDENTIFIER value="ed8c0a4e-7cf2-49df-bfed-fcfab0699ade_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="StatsJob" type="mapreduce"> |
|
| 11 |
<DESCRIPTION> |
|
| 12 |
map reduce job that exports hbase data and prepares them for import to the relation |
|
| 13 |
database used for statistics generation |
|
| 14 |
</DESCRIPTION> |
|
| 15 |
<STATIC_CONFIGURATION> |
|
| 16 |
|
|
| 15 | 17 |
<!-- I/O FORMAT --> |
| 16 | 18 |
<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" /> |
| 17 |
<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat" />
|
|
| 18 |
|
|
| 19 |
<!-- MAPPER -->
|
|
| 19 |
<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat" /> |
|
| 20 |
|
|
| 21 |
<!-- MAPPER -->
|
|
| 20 | 22 |
<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.statsExport.StatsMapper" /> |
| 21 | 23 |
<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" /> |
| 22 | 24 |
<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" /> |
| 23 | 25 |
|
| 24 |
<!--MultipleOutputs.addNamedOutput(job, type.toString(), TextOutputFormat.class, Text.class, Text.class);--> |
|
| 25 |
|
|
| 26 |
<!-- REDUCER --> |
|
| 26 |
<!-- REDUCER --> |
|
| 27 | 27 |
<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.statsExport.StatsReducer" /> |
| 28 |
<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text" />
|
|
| 29 |
<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat" />
|
|
| 28 |
<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text" /> |
|
| 29 |
<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat" /> |
|
| 30 | 30 |
|
| 31 | 31 |
<!-- MISC --> |
| 32 |
<PROPERTY key="mapred.output.compress" value="false" />
|
|
| 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" />
|
|
| 32 |
<PROPERTY key="mapred.output.compress" value="false" /> |
|
| 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 | 37 |
<PROPERTY key="mapreduce.reduce.speculative" value="false" /> |
| 38 |
|
|
| 39 |
<PROPERTY key="mapred.reduce.tasks" value="500" />
|
|
| 38 |
|
|
| 39 |
<PROPERTY key="mapred.reduce.tasks" value="10" />
|
|
| 40 | 40 |
<PROPERTY key="dfs.blocksize" value="32M" /> |
| 41 |
<!-- <PROPERTY key="user.name" value="dnet" /> --> |
|
| 42 | 41 |
|
| 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="hbase.mapred.inputtable" required="true" description="source hbase table" /> |
|
| 48 |
<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" /> |
|
| 49 |
<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs" /> |
|
| 50 |
|
|
| 51 |
<PARAM name="index.entity.links" required="true" description="entity joiner configuration" /> |
|
| 52 |
<PARAM name="contextmap" required="true" description="context map (ContextDSResources)" /> |
|
| 53 |
|
|
| 54 |
|
|
| 55 |
</JOB_INTERFACE> |
|
| 56 |
<SCAN> |
|
| 57 |
<FILTERS /> |
|
| 58 |
<FAMILIES /> |
|
| 59 |
</SCAN> |
|
| 60 |
</HADOOP_JOB> |
|
| 61 |
<STATUS> |
|
| 62 |
<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/> |
|
| 63 |
<RUNNING_INSTANCES value="0"/> |
|
| 64 |
<CUMULATIVE_RUN value="0" /> |
|
| 65 |
</STATUS> |
|
| 66 |
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS> |
|
| 67 |
</BODY> |
|
| 42 |
<!-- MULTIPLE OUTPUTS FOR REDUCER --> |
|
| 43 |
|
|
| 44 |
<PROPERTY key="mapreduce.multipleoutputs" value=" datasource" /> |
|
| 45 |
<PROPERTY key="mapreduce.multipleoutputs.namedOutput.datasource.format" value="TextOutputFormat.class" /> |
|
| 46 |
<PROPERTY key="mapreduce.multipleoutputs.namedOutput.datasource.key" value="Text.class" /> |
|
| 47 |
<PROPERTY key="mapreduce.multipleoutputs.namedOutput.datasource.value" value="Text.class" /> |
|
| 48 |
<!-- conf.setClass("mapreduce.multipleoutputs.namedOutput.datasource.format" outputFormatClass,
|
|
| 49 |
OutputFormat.class); conf.set(mapreduce.multipleoutputs, conf.get(mapreduce.multipleoutputs, |
|
| 50 |
"") + " " + namedOutput); MultipleOutputs.addNamedOutput(job, type.toString(), |
|
| 51 |
TextOutputFormat.class, Text.class, Text.class); --> |
|
| 52 |
</STATIC_CONFIGURATION> |
|
| 53 |
<JOB_INTERFACE> |
|
| 54 |
<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" /> |
|
| 55 |
<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" /> |
|
| 56 |
<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs" /> |
|
| 57 |
|
|
| 58 |
<PARAM name="mapred.output.delim" required="true" description="delim character used to seperate fields in hdfs dump files" /> |
|
| 59 |
<PARAM name="mapred.output.nullNum" required="true" description=" default string for Null Numeric Values" /> |
|
| 60 |
<PARAM name="mapred.output.nullString" required="true" description=" default string for Null String Values" /> |
|
| 61 |
|
|
| 62 |
<PARAM name="index.entity.links" required="true" description="entity joiner configuration" /> |
|
| 63 |
<PARAM name="contextmap" required="true" description="context map (ContextDSResources)" /> |
|
| 64 |
|
|
| 65 |
|
|
| 66 |
</JOB_INTERFACE> |
|
| 67 |
<SCAN> |
|
| 68 |
|
|
| 69 |
<FILTERS > |
|
| 70 |
|
|
| 71 |
</FILTERS> |
|
| 72 |
<FAMILIES> |
|
| 73 |
<FAMILY param="entityType" value="datasource"/> |
|
| 74 |
</FAMILIES> |
|
| 75 |
</SCAN> |
|
| 76 |
</HADOOP_JOB> |
|
| 77 |
<STATUS> |
|
| 78 |
<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/> |
|
| 79 |
<RUNNING_INSTANCES value="0"/> |
|
| 80 |
<CUMULATIVE_RUN value="0" /> |
|
| 81 |
</STATUS> |
|
| 82 |
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS> |
|
| 83 |
</BODY> |
|
| 68 | 84 |
</RESOURCE_PROFILE> |
Also available in: Unified diff
u