1 |
28797
|
claudio.at
|
<?xml version="1.0" encoding="UTF-8"?>
|
2 |
48139
|
alessia.ba
|
<RESOURCE_PROFILE>
|
3 |
28797
|
claudio.at
|
<HEADER>
|
4 |
48139
|
alessia.ba
|
<RESOURCE_IDENTIFIER value="0ef16778-4667-44b9-b453-3386f445beb1_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
|
5 |
|
|
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
|
6 |
|
|
<RESOURCE_KIND value="WorkflowDSResources"/>
|
7 |
|
|
<RESOURCE_URI value=""/>
|
8 |
|
|
<DATE_OF_CREATION value="2014-08-01T18:13:51.0Z"/>
|
9 |
28797
|
claudio.at
|
</HEADER>
|
10 |
|
|
<BODY>
|
11 |
|
|
<WORKFLOW_NAME>Index from HDFS</WORKFLOW_NAME>
|
12 |
|
|
<WORKFLOW_TYPE>Provision</WORKFLOW_TYPE>
|
13 |
|
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
14 |
|
|
<CONFIGURATION start="manual">
|
15 |
28892
|
claudio.at
|
<NODE name="setHdfsFile" type="SetHdfsFile" isStart="true">
|
16 |
48139
|
alessia.ba
|
<DESCRIPTION/>
|
17 |
28797
|
claudio.at
|
<PARAMETERS>
|
18 |
28898
|
claudio.at
|
<PARAM name="hdfsPath" type="string" managedBy="user" required="true">/tmp/indexrecords_db_openaireplus_node6_t_DMF.seq</PARAM>
|
19 |
|
|
<PARAM name="hdfsPathParam" type="string" managedBy="system" required="true">inputRecordsPath</PARAM>
|
20 |
28891
|
claudio.at
|
</PARAMETERS>
|
21 |
|
|
<ARCS>
|
22 |
48139
|
alessia.ba
|
<ARC to="prepare"/>
|
23 |
28891
|
claudio.at
|
</ARCS>
|
24 |
|
|
</NODE>
|
25 |
30038
|
claudio.at
|
<NODE name="findIndex" type="FindIndex" isStart="true">
|
26 |
48139
|
alessia.ba
|
<DESCRIPTION/>
|
27 |
28891
|
claudio.at
|
<PARAMETERS>
|
28 |
30038
|
claudio.at
|
<PARAM name="mdFormat" type="string" managedBy="user" required="true">DMF</PARAM>
|
29 |
|
|
<PARAM name="layout" type="string" managedBy="system" required="true">index</PARAM>
|
30 |
|
|
<PARAM name="interpretation" type="string" managedBy="system" required="true">openaire</PARAM>
|
31 |
28797
|
claudio.at
|
</PARAMETERS>
|
32 |
|
|
<ARCS>
|
33 |
48139
|
alessia.ba
|
<ARC name="found" to="waitIndex"/>
|
34 |
|
|
<ARC name="notFound" to="createIndex"/>
|
35 |
30038
|
claudio.at
|
</ARCS>
|
36 |
|
|
</NODE>
|
37 |
|
|
<NODE name="createIndex" type="CreateIndex">
|
38 |
48139
|
alessia.ba
|
<DESCRIPTION/>
|
39 |
|
|
<PARAMETERS/>
|
40 |
30038
|
claudio.at
|
<ARCS>
|
41 |
48139
|
alessia.ba
|
<ARC to="waitIndex"/>
|
42 |
30038
|
claudio.at
|
</ARCS>
|
43 |
|
|
</NODE>
|
44 |
|
|
<NODE name="waitIndex">
|
45 |
48139
|
alessia.ba
|
<DESCRIPTION/>
|
46 |
|
|
<PARAMETERS/>
|
47 |
30038
|
claudio.at
|
<ARCS>
|
48 |
48139
|
alessia.ba
|
<ARC to="prepare"/>
|
49 |
28797
|
claudio.at
|
</ARCS>
|
50 |
30038
|
claudio.at
|
</NODE>
|
51 |
28891
|
claudio.at
|
<NODE name="prepare" type="PrepareIndexJob" isJoin="true">
|
52 |
28797
|
claudio.at
|
<DESCRIPTION>Prepare indexing</DESCRIPTION>
|
53 |
|
|
<PARAMETERS>
|
54 |
48139
|
alessia.ba
|
<PARAM name="rottenRecordsPathParam" type="string" managedBy="system" required="true">rottenRecordsPath</PARAM>
|
55 |
28797
|
claudio.at
|
<PARAM name="layoutToRecordStylesheet" type="string" managedBy="system" required="true">/eu/dnetlib/msro/openaireplus/workflows/index/openaireLayoutToRecordStylesheet.xsl</PARAM>
|
56 |
48139
|
alessia.ba
|
<PARAM name="oafSchemaLocationProperty" type="string" managedBy="system" required="true">oaf.schema.location</PARAM>
|
57 |
28797
|
claudio.at
|
</PARAMETERS>
|
58 |
|
|
<ARCS>
|
59 |
48139
|
alessia.ba
|
<ARC to="cleanupRotten"/>
|
60 |
28797
|
claudio.at
|
</ARCS>
|
61 |
48139
|
alessia.ba
|
</NODE>
|
62 |
28797
|
claudio.at
|
<NODE name="cleanupRotten" type="DeleteHdfsPathJob">
|
63 |
|
|
<DESCRIPTION>hdfs cleanup (rotten)</DESCRIPTION>
|
64 |
|
|
<PARAMETERS>
|
65 |
|
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
66 |
|
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
67 |
48139
|
alessia.ba
|
{
|
68 |
|
|
'path' : 'rottenRecordsPath'
|
69 |
28797
|
claudio.at
|
}
|
70 |
48139
|
alessia.ba
|
</PARAM>
|
71 |
28797
|
claudio.at
|
</PARAMETERS>
|
72 |
|
|
<ARCS>
|
73 |
48139
|
alessia.ba
|
<ARC to="updateIndex"/>
|
74 |
28797
|
claudio.at
|
</ARCS>
|
75 |
48139
|
alessia.ba
|
</NODE>
|
76 |
28797
|
claudio.at
|
<NODE name="updateIndex" type="SubmitHadoopJob">
|
77 |
|
|
<DESCRIPTION>M/O index records</DESCRIPTION>
|
78 |
|
|
<PARAMETERS>
|
79 |
|
|
<PARAM required="true" type="string" name="cluster" managedBy="system">DM</PARAM>
|
80 |
48139
|
alessia.ba
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">indexFeedJob</PARAM>
|
81 |
28797
|
claudio.at
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
82 |
48139
|
alessia.ba
|
{
|
83 |
|
|
'mapred.input.dir' : 'inputRecordsPath',
|
84 |
|
|
'mapred.output.dir' : 'rottenRecordsPath',
|
85 |
|
|
'index.xslt' : 'index.xslt',
|
86 |
|
|
'index.solr.url' : 'index.solr.url',
|
87 |
|
|
'index.solr.collection' : 'index.solr.collection',
|
88 |
|
|
'index.buffer.flush.threshold' : 'index.buffer.flush.threshold',
|
89 |
|
|
'index.shutdown.wait.time' : 'index.shutdown.wait.time',
|
90 |
|
|
'index.solr.sim.mode' : 'index.solr.sim.mode',
|
91 |
|
|
'index.feed.timestamp' : 'index.feed.timestamp'
|
92 |
28797
|
claudio.at
|
}
|
93 |
|
|
</PARAM>
|
94 |
|
|
</PARAMETERS>
|
95 |
|
|
<ARCS>
|
96 |
48139
|
alessia.ba
|
<ARC to="finalize"/>
|
97 |
28797
|
claudio.at
|
</ARCS>
|
98 |
|
|
</NODE>
|
99 |
|
|
<NODE name="finalize" type="FinalizeIndexFeeding">
|
100 |
|
|
<DESCRIPTION>commit changes</DESCRIPTION>
|
101 |
48139
|
alessia.ba
|
<PARAMETERS/>
|
102 |
28797
|
claudio.at
|
<ARCS>
|
103 |
48139
|
alessia.ba
|
<ARC to="success"/>
|
104 |
28797
|
claudio.at
|
</ARCS>
|
105 |
|
|
</NODE>
|
106 |
|
|
</CONFIGURATION>
|
107 |
48139
|
alessia.ba
|
<STATUS/>
|
108 |
28797
|
claudio.at
|
</BODY>
|
109 |
|
|
</RESOURCE_PROFILE>
|
110 |
|
|
|