1 |
58329
|
claudio.at
|
<RESOURCE_PROFILE>
|
2 |
|
|
<HEADER>
|
3 |
|
|
<RESOURCE_IDENTIFIER value="2483c5e4-dde8-4452-bd9c-38d436d8d941_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
|
4 |
|
|
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
|
5 |
|
|
<RESOURCE_KIND value="WorkflowDSResources"/>
|
6 |
|
|
<RESOURCE_URI value=""/>
|
7 |
58349
|
claudio.at
|
<DATE_OF_CREATION value="2020-03-24T11:15:39+00:00"/>
|
8 |
58329
|
claudio.at
|
</HEADER>
|
9 |
|
|
<BODY>
|
10 |
|
|
<WORKFLOW_NAME>Offline Deduplication [OCEAN]</WORKFLOW_NAME>
|
11 |
|
|
<WORKFLOW_TYPE>Deduplication</WORKFLOW_TYPE>
|
12 |
|
|
<WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
|
13 |
|
|
<CONFIGURATION start="manual">
|
14 |
|
|
|
15 |
|
|
<NODE name="setGraphBasePath" type="SetEnvParameter" isStart="true">
|
16 |
|
|
<DESCRIPTION>Set the target directory to materialize the graph</DESCRIPTION>
|
17 |
|
|
<PARAMETERS>
|
18 |
|
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">graphBasePath</PARAM>
|
19 |
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/var/lib/dnet/graph_BETA</PARAM>
|
20 |
|
|
</PARAMETERS>
|
21 |
|
|
<ARCS>
|
22 |
|
|
<ARC to="REUSE_RAW_GRAPH"/>
|
23 |
|
|
</ARCS>
|
24 |
|
|
</NODE>
|
25 |
|
|
|
26 |
|
|
<NODE name="REUSE_RAW_GRAPH" type="Selection">
|
27 |
|
|
<DESCRIPTION>Is the raw graph already available?</DESCRIPTION>
|
28 |
|
|
<PARAMETERS>
|
29 |
|
|
<PARAM function="validValues(['YES', 'NO'])" managedBy="user" name="selection" required="true" type="string">YES</PARAM>
|
30 |
|
|
</PARAMETERS>
|
31 |
|
|
<ARCS>
|
32 |
58350
|
claudio.at
|
<ARC name="YES" to="setRawGraphPath"/>
|
33 |
58329
|
claudio.at
|
<ARC name="NO" to="rawGraph"/>
|
34 |
|
|
</ARCS>
|
35 |
|
|
</NODE>
|
36 |
|
|
|
37 |
|
|
<NODE name="rawGraph" type="SubmitHadoopJob">
|
38 |
|
|
<DESCRIPTION>create the raw graph</DESCRIPTION>
|
39 |
|
|
<PARAMETERS>
|
40 |
|
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">executeOozieJob</PARAM>
|
41 |
|
|
<PARAM required="true" type="string" name="cluster" managedBy="system">IIS</PARAM>
|
42 |
|
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
43 |
|
|
{
|
44 |
|
|
'graphBasePath' : 'graphBasePath'
|
45 |
|
|
}
|
46 |
|
|
</PARAM>
|
47 |
|
|
<PARAM required="true" type="string" name="params" managedBy="system">
|
48 |
|
|
{
|
49 |
|
|
'oozie.wf.application.path' : '/lib/dnet/migration/wfs/regular_all_steps/oozie_app',
|
50 |
|
|
'mongoURL' : 'mongodb://beta.services.openaire.eu',
|
51 |
|
|
'mongoDb' : 'mdstore',
|
52 |
|
|
'postgresURL' : 'jdbc:postgresql://beta.services.openaire.eu:5432/dnet_openaireplus',
|
53 |
|
|
'postgresUser' : 'dnet',
|
54 |
58349
|
claudio.at
|
'postgresPassword' : 'xxx',
|
55 |
58329
|
claudio.at
|
'workingPath' : '/tmp/dhp_migration_beta',
|
56 |
|
|
'reuseContent' : 'true'
|
57 |
|
|
}
|
58 |
|
|
</PARAM>
|
59 |
|
|
<PARAM required="true" type="boolean" name="simulation" managedBy="system">false</PARAM>
|
60 |
|
|
<PARAM required="true" type="string" name="oozieReportActionsCsv" managedBy="system">build-report</PARAM>
|
61 |
|
|
</PARAMETERS>
|
62 |
|
|
<ARCS>
|
63 |
58349
|
claudio.at
|
<ARC to="setRawGraphPath"/>
|
64 |
58329
|
claudio.at
|
</ARCS>
|
65 |
|
|
</NODE>
|
66 |
|
|
|
67 |
58349
|
claudio.at
|
<NODE name="setRawGraphPath" type="SetEnvParameter">
|
68 |
|
|
<DESCRIPTION>Set the path containing the raw graph</DESCRIPTION>
|
69 |
58329
|
claudio.at
|
<PARAMETERS>
|
70 |
58349
|
claudio.at
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">rawGraphPath</PARAM>
|
71 |
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/var/lib/dnet/graph_BETA/graph_raw</PARAM>
|
72 |
58329
|
claudio.at
|
</PARAMETERS>
|
73 |
|
|
<ARCS>
|
74 |
58349
|
claudio.at
|
<ARC to="setDedupWorkingPath"/>
|
75 |
58329
|
claudio.at
|
</ARCS>
|
76 |
|
|
</NODE>
|
77 |
|
|
|
78 |
58349
|
claudio.at
|
<NODE name="setDedupWorkingPath" type="SetEnvParameter">
|
79 |
|
|
<DESCRIPTION>Set the path used by dedup workflows to store intermediate data</DESCRIPTION>
|
80 |
58329
|
claudio.at
|
<PARAMETERS>
|
81 |
58349
|
claudio.at
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">dedupWorkingPath</PARAM>
|
82 |
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/tmp/beta_dedup</PARAM>
|
83 |
58329
|
claudio.at
|
</PARAMETERS>
|
84 |
|
|
<ARCS>
|
85 |
58349
|
claudio.at
|
<ARC to="setDedupGraphPath"/>
|
86 |
58329
|
claudio.at
|
</ARCS>
|
87 |
|
|
</NODE>
|
88 |
|
|
|
89 |
58349
|
claudio.at
|
<NODE name="setDedupGraphPath" type="SetEnvParameter">
|
90 |
|
|
<DESCRIPTION>Set the target path to store the deduped graph</DESCRIPTION>
|
91 |
58329
|
claudio.at
|
<PARAMETERS>
|
92 |
58349
|
claudio.at
|
<PARAM managedBy="system" name="parameterName" required="true" type="string">dedupGraphPath</PARAM>
|
93 |
|
|
<PARAM managedBy="user" name="parameterValue" required="true" type="string">/var/lib/dnet/graph_BETA/graph_dedup</PARAM>
|
94 |
|
|
</PARAMETERS>
|
95 |
|
|
<ARCS>
|
96 |
|
|
<ARC to="setDedupConfig"/>
|
97 |
|
|
</ARCS>
|
98 |
|
|
</NODE>
|
99 |
|
|
|
100 |
|
|
<NODE name="setDedupConfig" type="SetDedupConfiguration">
|
101 |
|
|
<DESCRIPTION>Set Dedup conf</DESCRIPTION>
|
102 |
|
|
<PARAMETERS>
|
103 |
|
|
<PARAM function="obtainValues('dedupOrchestrations', {})" managedBy="user" name="dedupConfigSequence" required="true" type="string">dedup-similarity-organization</PARAM>
|
104 |
58329
|
claudio.at
|
<PARAM managedBy="system" name="dedupConfigSequenceParam" required="true" type="string">dedup.conf.queue</PARAM>
|
105 |
|
|
</PARAMETERS>
|
106 |
|
|
<ARCS>
|
107 |
|
|
<ARC to="duplicateScan"/>
|
108 |
|
|
</ARCS>
|
109 |
|
|
</NODE>
|
110 |
|
|
|
111 |
|
|
<NODE name="duplicateScan" type="SubmitHadoopJob">
|
112 |
|
|
<DESCRIPTION>create the raw graph</DESCRIPTION>
|
113 |
|
|
<PARAMETERS>
|
114 |
|
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">executeOozieJob</PARAM>
|
115 |
|
|
<PARAM required="true" type="string" name="cluster" managedBy="system">IIS</PARAM>
|
116 |
|
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
117 |
|
|
{
|
118 |
58349
|
claudio.at
|
'actionSetId' : 'setActionSetId',
|
119 |
|
|
'graphBasePath' : 'rawGraphPath',
|
120 |
|
|
'workingPath' : 'dedupWorkingPath',
|
121 |
|
|
'dedupGraphPath': 'dedupGraphPath'
|
122 |
58329
|
claudio.at
|
}
|
123 |
|
|
</PARAM>
|
124 |
|
|
<PARAM required="true" type="string" name="params" managedBy="system">
|
125 |
|
|
{
|
126 |
58349
|
claudio.at
|
'oozie.wf.application.path' : '/lib/dnet/dedup/scan/oozie_app'
|
127 |
58329
|
claudio.at
|
}
|
128 |
|
|
</PARAM>
|
129 |
|
|
<PARAM required="true" type="string" name="oozieReportActionsCsv" managedBy="system">build-report</PARAM>
|
130 |
|
|
</PARAMETERS>
|
131 |
|
|
<ARCS>
|
132 |
58349
|
claudio.at
|
<ARC to="dedupConsistency"/>
|
133 |
58329
|
claudio.at
|
</ARCS>
|
134 |
|
|
</NODE>
|
135 |
|
|
|
136 |
58349
|
claudio.at
|
<NODE name="dedupConsistency" type="SubmitHadoopJob">
|
137 |
|
|
<DESCRIPTION>mark duplicates as deleted and redistribute the relationships</DESCRIPTION>
|
138 |
|
|
<PARAMETERS>
|
139 |
|
|
<PARAM required="true" type="string" name="hadoopJob" managedBy="system">executeOozieJob</PARAM>
|
140 |
|
|
<PARAM required="true" type="string" name="cluster" managedBy="system">IIS</PARAM>
|
141 |
|
|
<PARAM required="true" type="string" name="envParams" managedBy="system">
|
142 |
|
|
{
|
143 |
|
|
'graphBasePath' : 'rawGraphPath',
|
144 |
|
|
'dedupGraphPath': 'dedupGraphPath'
|
145 |
|
|
'workingPath' : 'dedupWorkingPath',
|
146 |
|
|
}
|
147 |
|
|
</PARAM>
|
148 |
|
|
<PARAM required="true" type="string" name="params" managedBy="system">
|
149 |
|
|
{
|
150 |
|
|
'oozie.wf.application.path' : '/lib/dnet/dedup/consistency/oozie_app'
|
151 |
|
|
}
|
152 |
|
|
</PARAM>
|
153 |
|
|
<PARAM required="true" type="string" name="oozieReportActionsCsv" managedBy="system">build-report</PARAM>
|
154 |
|
|
</PARAMETERS>
|
155 |
58329
|
claudio.at
|
<ARCS>
|
156 |
|
|
<ARC to="success"/>
|
157 |
|
|
</ARCS>
|
158 |
|
|
</NODE>
|
159 |
|
|
|
160 |
58349
|
claudio.at
|
|
161 |
58329
|
claudio.at
|
</CONFIGURATION>
|
162 |
|
|
<STATUS>
|
163 |
|
|
<LAST_EXECUTION_ID>wf_20200320_224915_420</LAST_EXECUTION_ID>
|
164 |
|
|
<LAST_EXECUTION_DATE>2020-03-20T23:07:47+00:00</LAST_EXECUTION_DATE>
|
165 |
|
|
<LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
|
166 |
|
|
<LAST_EXECUTION_ERROR/>
|
167 |
|
|
<LAST_EXECUTION_OUTPUT name="mainlog:storeSimilarities:count">0</LAST_EXECUTION_OUTPUT>
|
168 |
|
|
<LAST_EXECUTION_OUTPUT name="mainlog:storeDissimilarities:count">0</LAST_EXECUTION_OUTPUT>
|
169 |
|
|
</STATUS>
|
170 |
|
|
</BODY>
|
171 |
|
|
</RESOURCE_PROFILE>
|