|
1
|
<RESOURCE_PROFILE>
|
|
2
|
<HEADER>
|
|
3
|
<RESOURCE_IDENTIFIER value="1c34963b-75b3-4440-9f42-72445a26c077_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="indexFeedJob" type="mapreduce">
|
|
11
|
<DESCRIPTION>map reduce job that joins the entities on the hbase table and produces a sequence file containig the xml records</DESCRIPTION>
|
|
12
|
<STATIC_CONFIGURATION>
|
|
13
|
|
|
14
|
<!-- I/O FORMAT -->
|
|
15
|
<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat" />
|
|
16
|
<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat" />
|
|
17
|
|
|
18
|
<!-- MAPPER -->
|
|
19
|
<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.index.IndexFeedMapper" />
|
|
20
|
<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
|
|
21
|
<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text" />
|
|
22
|
|
|
23
|
<!-- JOB GLOBAL -->
|
|
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.map.tasks.speculative.execution" value="false" />
|
|
29
|
<PROPERTY key="mapreduce.map.speculative" value="false" />
|
|
30
|
<PROPERTY key="mapred.reduce.tasks" value="0" />
|
|
31
|
<!-- <PROPERTY key="user.name" value="dnet" /> -->
|
|
32
|
|
|
33
|
<!-- Uncomment to override the default lib path -->
|
|
34
|
<!-- <PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
|
|
35
|
</STATIC_CONFIGURATION>
|
|
36
|
<JOB_INTERFACE>
|
|
37
|
<PARAM name="mapred.input.dir" required="true" description="source sequence file on hdfs" />
|
|
38
|
<PARAM name="mapred.output.dir" required="true" description="destination path on hdfs for rotten index xml records" />
|
|
39
|
|
|
40
|
<PARAM name="index.solr.url.local" required="false" description="solr url used to feed the records back to a single index instance" />
|
|
41
|
<PARAM name="index.solr.url.list" required="true" description="solr url list used to feed the records on the sharded solr cluster" />
|
|
42
|
<PARAM name="index.solr.url.zk" required="true" description="zookeeper solr quorum address list" />
|
|
43
|
<PARAM name="index.solr.collection" required="true" description="target solr collection to be fed" />
|
|
44
|
<PARAM name="index.solr.local.feeding" required="true" description="instructs about the index feeding mode" />
|
|
45
|
|
|
46
|
<PARAM name="id" required="true" description="index DS id" />
|
|
47
|
<PARAM name="index.shutdown.wait.time" required="true" description="wait time before shut down the solr client pool" />
|
|
48
|
<PARAM name="index.buffer.flush.threshold" required="true" description="indexing buffer flush threshold" />
|
|
49
|
<PARAM name="index.feed.timestamp" required="true" description="timestamp used as ds_version" />
|
|
50
|
<PARAM name="index.solr.sim.mode" required="true" description="boolean value, allows to run this job in simulation mode" />
|
|
51
|
<PARAM name="index.xslt" required="true" description="record transformer created by the MSRO service" />
|
|
52
|
</JOB_INTERFACE>
|
|
53
|
<SCAN>
|
|
54
|
<FILTERS />
|
|
55
|
<FAMILIES />
|
|
56
|
</SCAN>
|
|
57
|
</HADOOP_JOB>
|
|
58
|
<STATUS>
|
|
59
|
<LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
|
|
60
|
<RUNNING_INSTANCES value="0"/>
|
|
61
|
<CUMULATIVE_RUN value="0" />
|
|
62
|
</STATUS>
|
|
63
|
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
|
|
64
|
</BODY>
|
|
65
|
</RESOURCE_PROFILE>
|