1
|
<?xml version="1.0" encoding="UTF-8"?>
|
2
|
<RESOURCE_PROFILE>
|
3
|
<HEADER>
|
4
|
<RESOURCE_IDENTIFIER value="732656e3-5ac6-4344-9d1f-f5c805f53a06_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
|
5
|
<RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
|
6
|
<RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
|
7
|
<RESOURCE_URI value=""/>
|
8
|
<DATE_OF_CREATION value="2019-04-11T11:31:58+02:00"/>
|
9
|
</HEADER>
|
10
|
<BODY>
|
11
|
<HADOOP_JOB name="dnetHadoopCollection" type="oozie">
|
12
|
<DESCRIPTION>IIS preprocessing</DESCRIPTION>
|
13
|
<STATIC_CONFIGURATION><!-- Cluster wide -->
|
14
|
<PROPERTY key="queueName" value="default"/>
|
15
|
<PROPERTY key="user.name" value="dnet"/><!-- Runtime -->
|
16
|
<PROPERTY value="/user/dnet/oozieApp/collection" key="oozie.wf.application.path"/>
|
17
|
<PROPERTY value="false" key="oozie.wf.validate.ForkJoin"/>
|
18
|
<PROPERTY key="oozie.use.system.libpath" value="True"/>
|
19
|
<PROPERTY value="False" key="security_enabled"/>
|
20
|
<PROPERTY key="dryrun" value="True"/>
|
21
|
<PROPERTY key="oozie.action.sharelib.for.spark" value="spark2"/>
|
22
|
<PROPERTY key="metadataEncoding" value="XML"/>
|
23
|
</STATIC_CONFIGURATION>
|
24
|
<JOB_INTERFACE>
|
25
|
<PARAM name="workflowId" required="true" description="the Dnet Workflow Identifier"/>
|
26
|
<PARAM required="true" description="the path to store the sequence file of the native metadata collected" name="sequenceFilePath"/><!-- value="/user/sandro.labruzzo/mdstores/oai_1"-->
|
27
|
<PARAM description="the path of the native mdstore" required="true" name="mdStorePath"/>
|
28
|
<PARAM description="A json encoding of the API Description class" name="apiDescription" required="true"/>
|
29
|
<PARAM required="true" name="dataSourceInfo" description="A json encoding of the Datasource Info"/>
|
30
|
<PARAM description="An xpath to retrieve the metadata idnentifier for the generation of DNet Identifier " name="identifierPath" required="true"/>
|
31
|
<PARAM name="timestamp" description="The timestamp of the collection date" required="true"/>
|
32
|
</JOB_INTERFACE>
|
33
|
</HADOOP_JOB>
|
34
|
<STATUS>
|
35
|
<LAST_SUBMISSION_DATE value="2019-04-11T15:33:02+02:00"/>
|
36
|
<RUNNING_INSTANCES value="2"/>
|
37
|
<CUMULATIVE_RUN value="64"/>
|
38
|
</STATUS>
|
39
|
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
|
40
|
</BODY>
|
41
|
</RESOURCE_PROFILE>
|