1
|
<RESOURCE_PROFILE>
|
2
|
<HEADER>
|
3
|
<RESOURCE_IDENTIFIER value="d730a831-a982-4034-a890-de98fd972e87_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
|
4
|
<RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
|
5
|
<RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
|
6
|
<RESOURCE_URI value=""/>
|
7
|
<DATE_OF_CREATION value="2017-07-05T15:05:50+00:00"/>
|
8
|
</HEADER>
|
9
|
<BODY>
|
10
|
<HADOOP_JOB name="iisMainJobV2" type="oozie">
|
11
|
<DESCRIPTION>IIS main workflow</DESCRIPTION>
|
12
|
<STATIC_CONFIGURATION>
|
13
|
|
14
|
<!-- Cluster wide -->
|
15
|
<PROPERTY key="user.name" value="dnet.production"/>
|
16
|
|
17
|
<!-- Runtime -->
|
18
|
<PROPERTY key="metadataextraction_default_cache_location" value="/user/${user.name}/iis/cache/metadataextraction"/>
|
19
|
<!-- skipping _default chunk from property name since 2017.02.21 -->
|
20
|
<PROPERTY key="metadataextraction_cache_location" value="/user/${user.name}/iis/cache/metadataextraction"/>
|
21
|
<PROPERTY key="ingest_pmc_cache_location" value="/user/${user.name}/iis/cache/ingestpmc"/>
|
22
|
<PROPERTY key="software_webcrawl_cache_location" value="/user/${user.name}/iis/cache/webcrawler"/>
|
23
|
<PROPERTY key="import_content_objectstore_s3_keystore_location" value="jceks://hdfs/user/${user.name}/secrets/s3keyfile.jceks"/>
|
24
|
<PROPERTY key="export_documentssimilarity_threshold" value="0.7"/>
|
25
|
<PROPERTY key="export_trust_level_threshold_document_classes" value="0.5"/>
|
26
|
<PROPERTY key="export_trust_level_threshold_document_referencedDatasets" value="0.0"/>
|
27
|
<PROPERTY key="export_trust_level_threshold_document_referencedProjects" value="0.0"/>
|
28
|
<PROPERTY key="export_trust_level_threshold_document_pdb" value="0.0"/>
|
29
|
<PROPERTY key="export_trust_level_threshold_document_software_url" value="0.0"/>
|
30
|
<PROPERTY key="export_trust_level_threshold_matched_doc_organizations" value="0.0"/>
|
31
|
<PROPERTY key="output_remote_location" value="webhdfs://namenode.hadoop.dm.openaire.eu:50071/var/lib/dnet/actionManager_PROD"/>
|
32
|
<PROPERTY key="reports_external_path" value="/user/${user.name}/iis/reports/${execution_environment}"/>
|
33
|
<PROPERTY key="import_hbase_dump_location" value="webhdfs://namenode.hadoop.dm.openaire.eu:50071/tmp/db_openaireplus_services.export.2017.07.19"/>
|
34
|
<PROPERTY key="oozieServiceLoc" value="http://iis-cdh5-test-m3:11000/oozie"/>
|
35
|
</STATIC_CONFIGURATION>
|
36
|
<JOB_INTERFACE>
|
37
|
<PARAM description="oozie job application absolute path" name="oozie.wf.application.path" required="true"/>
|
38
|
<PARAM description="objectStore service endpoint" name="import_content_object_store_location" required="true"/>
|
39
|
<PARAM description="csv list of the available object stores subject to processing" name="import_content_objectstores_csv" required="true"/>
|
40
|
<PARAM description="mdstore service location" name="import_mdstore_service_location" required="true"/>
|
41
|
<PARAM description="mdstore ids for dataset records" name="import_dataset_mdstore_ids_csv" required="true"/>
|
42
|
<PARAM description="target action set for referenced datasets" name="export_action_set_id_document_referencedDatasets" required="true"/>
|
43
|
<PARAM description="target action set for referenced projects" name="export_action_set_id_document_referencedProjects" required="true"/>
|
44
|
<PARAM description="target action set for research initiative outcome" name="export_action_set_id_document_research_initiative" required="true"/>
|
45
|
<PARAM description="target action set for document similarities" name="export_action_set_id_document_similarities_standard" required="true"/>
|
46
|
<PARAM description="target action set for document classification" name="export_action_set_id_document_classes" required="true"/>
|
47
|
<PARAM description="target action set for document citations" name="export_action_set_id_document_referencedDocuments" required="true"/>
|
48
|
<PARAM description="target action set for dataset entities" name="export_action_set_id_entity_dataset" required="true"/>
|
49
|
<PARAM description="target action set for pdb entities" name="export_action_set_id_document_pdb" required="true"/>
|
50
|
<PARAM description="software reference extraction actionset identifier" name="export_action_set_id_document_software_url" required="true"/>
|
51
|
<PARAM description="target action set for affiliations" name="export_action_set_id_matched_doc_organizations" required="true"/>
|
52
|
|
53
|
|
54
|
<!-- flags to enable/disable IIS modules -->
|
55
|
<PARAM description="enable/disable the citation matching module" name="active_citationmatching" required="true"/>
|
56
|
<PARAM description="enable/disable the document classification module" name="active_documentsclassification" required="true"/>
|
57
|
<PARAM description="enable/disable the document similarity module" name="active_documentssimilarity" required="true"/>
|
58
|
<PARAM description="enable/disable the dataset reference extraction module" name="active_referenceextraction_dataset" required="true"/>
|
59
|
<PARAM description="enable/disable the project reference extracion module" name="active_referenceextraction_project" required="true"/>
|
60
|
<PARAM description="enable/disable the research initiative extraction module" name="active_referenceextraction_researchinitiative" required="true"/>
|
61
|
<PARAM description="enable/disable the protein data bank extraction module" name="active_referenceextraction_pdb" required="true"/>
|
62
|
<PARAM description="enable/disable the software reference extraction algorithm" name="active_referenceextraction_software_url" required="true"/>
|
63
|
</JOB_INTERFACE>
|
64
|
</HADOOP_JOB>
|
65
|
<STATUS>
|
66
|
<LAST_SUBMISSION_DATE value="2017-07-02T09:39:51+00:00"/>
|
67
|
<RUNNING_INSTANCES value="0"/>
|
68
|
<CUMULATIVE_RUN value="13"/>
|
69
|
</STATUS>
|
70
|
<SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
|
71
|
</BODY>
|
72
|
</RESOURCE_PROFILE>
|
73
|
|