1 |
44367
|
claudio.at
|
<RESOURCE_PROFILE>
|
2 |
|
|
<HEADER>
|
3 |
|
|
<RESOURCE_IDENTIFIER value="b6addffa-bae1-42db-811c-ae20a5c514a6_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
|
4 |
|
|
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
|
5 |
|
|
<RESOURCE_KIND value="WorkflowDSResources"/>
|
6 |
|
|
<RESOURCE_URI value=""/>
|
7 |
|
|
<DATE_OF_CREATION value="2006-05-04T18:13:51.0Z"/>
|
8 |
|
|
</HEADER>
|
9 |
|
|
<BODY>
|
10 |
|
|
<WORKFLOW_NAME menuSection="InfoSpace Provision">DLI Data Provision</WORKFLOW_NAME>
|
11 |
|
|
<WORKFLOW_DESCRIPTION>DLI Data Provision</WORKFLOW_DESCRIPTION>
|
12 |
|
|
<WORKFLOW_INFO/>
|
13 |
|
|
<WORKFLOW_FAMILY>Data Provision</WORKFLOW_FAMILY>
|
14 |
|
|
<WORKFLOW_PRIORITY>35</WORKFLOW_PRIORITY>
|
15 |
|
|
<CONFIGURATION status="EXECUTABLE" start="MANUAL">
|
16 |
|
|
|
17 |
|
|
<PARAMETERS>
|
18 |
46058
|
sandro.lab
|
<PARAM name="table" description="HBase table to be dropped" required="true" managedBy="user"
|
19 |
|
|
type="string"/>
|
20 |
|
|
<PARAM name="cluster" description="Hadoop cluster logical name" required="true" managedBy="user"
|
21 |
|
|
type="string" function="validValues(['DM','IIS'])"/>
|
22 |
44367
|
claudio.at
|
<PARAM name="reuseRegionInfo" description="Reuse table splits" required="true" managedBy="user" type="boolean">true</PARAM>
|
23 |
|
|
|
24 |
45704
|
claudio.at
|
<PARAM name="mappingPublication" description="xslt mapping for publications" required="true" type="string" managedBy="user" function="listProfiles('TransformationRuleDSResourceType', '//TITLE', 'DLI2HBASE:')"/>
|
25 |
44367
|
claudio.at
|
<PARAM name="reusePublication" description="reuse publications on HDFS?" required="true" type="boolean" managedBy="user"/>
|
26 |
|
|
<PARAM name="hdfsPathPublication" description="hdfs path for publications" required="true" type="string" managedBy="user"/>
|
27 |
|
|
|
28 |
45704
|
claudio.at
|
<PARAM name="reuseResolvedPublication" description="reuse resolved publications on HDFS?" required="true" type="boolean" managedBy="user"/>
|
29 |
|
|
<PARAM name="hdfsPathResolvedPublication" description="hdfs path for resolved publications" required="true" type="string" managedBy="user"/>
|
30 |
|
|
|
31 |
|
|
|
32 |
|
|
<PARAM name="mappingDataset" description="xslt mapping for datasets" required="true" type="string" managedBy="user" function="listProfiles('TransformationRuleDSResourceType', '//TITLE', 'DLI2HBASE:')"/>
|
33 |
44367
|
claudio.at
|
<PARAM name="reuseDataset" description="reuse datasets on HDFS?" required="true" type="boolean" managedBy="user"/>
|
34 |
|
|
<PARAM name="hdfsPathDataset" description="hdfs path for datasets" required="true" type="string" managedBy="user"/>
|
35 |
|
|
|
36 |
45704
|
claudio.at
|
<PARAM name="reuseResolvedDataset" description="reuse resolved datasets on HDFS?" required="true" type="boolean" managedBy="user"/>
|
37 |
|
|
<PARAM name="hdfsPathResolvedDataset" description="hdfs path for resolved datasets" required="true" type="string" managedBy="user"/>
|
38 |
|
|
|
39 |
|
|
|
40 |
|
|
<PARAM name="mappingUnresolved" description="xslt mapping for unresolved objects" required="true" type="string" managedBy="user" function="listProfiles('TransformationRuleDSResourceType', '//TITLE', 'DLI2HBASE:')"/>
|
41 |
|
|
<PARAM name="reuseUnresolved" description="reuse unresolved objects on HDFS?" required="true" type="boolean" managedBy="user"/>
|
42 |
|
|
<PARAM name="hdfsPathUnresolved" description="hdfs path for unresolved objects" required="true" type="string" managedBy="user"/>
|
43 |
|
|
|
44 |
|
|
<PARAM name="actionSets" description="csv list of action sets to be promoted" required="true" type="string" managedBy="user"/>
|
45 |
46058
|
sandro.lab
|
<PARAM name="minDistWorkDir" description="work directory for the minDist algorithm" required="true"
|
46 |
|
|
type="string" managedBy="user"/>
|
47 |
|
|
<PARAM description="sequence of entity to be processed" managedBy="user" name="entitysequence"
|
48 |
|
|
required="true" type="string">dataset,publication,unknown
|
49 |
|
|
</PARAM>
|
50 |
45704
|
claudio.at
|
|
51 |
44367
|
claudio.at
|
|
52 |
|
|
</PARAMETERS>
|
53 |
|
|
<WORKFLOW>
|
54 |
|
|
|
55 |
|
|
<NODE name="resetHbase" type="LaunchWorkflowTemplate" isStart="true">
|
56 |
|
|
<DESCRIPTION>Reset HBase table</DESCRIPTION>
|
57 |
|
|
<PARAMETERS>
|
58 |
|
|
<PARAM name="wfTemplateId" value="75345aba-c069-43f4-90aa-e13688d9845e_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ=="/>
|
59 |
|
|
<PARAM name="wfTemplateParams">
|
60 |
|
|
<MAP>
|
61 |
|
|
<ENTRY key="table" ref="table"/>
|
62 |
|
|
<ENTRY key="cluster" ref="cluster"/>
|
63 |
|
|
<ENTRY key="reuseRegionInfo" ref="reuseRegionInfo"/>
|
64 |
|
|
</MAP>
|
65 |
|
|
</PARAM>
|
66 |
|
|
</PARAMETERS>
|
67 |
|
|
<ARCS>
|
68 |
45704
|
claudio.at
|
<ARC to="pmf2hbase"/>
|
69 |
44367
|
claudio.at
|
</ARCS>
|
70 |
|
|
</NODE>
|
71 |
|
|
|
72 |
45704
|
claudio.at
|
|
73 |
|
|
|
74 |
|
|
<NODE name="pmf2hbase" type="LaunchWorkflowTemplate">
|
75 |
|
|
<DESCRIPTION>PMF Publications to HBase</DESCRIPTION>
|
76 |
44367
|
claudio.at
|
<PARAMETERS>
|
77 |
45704
|
claudio.at
|
<PARAM name="wfTemplateId" value="5b05a65a-4eeb-4862-bc55-b35c7ec3baf0_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ=="/>
|
78 |
44367
|
claudio.at
|
<PARAM name="wfTemplateParams">
|
79 |
|
|
<MAP>
|
80 |
|
|
<ENTRY key="hbaseTable" ref="table"/>
|
81 |
|
|
<ENTRY key="cluster" ref="cluster"/>
|
82 |
45704
|
claudio.at
|
<ENTRY key="mdFormat" value="PMF"/>
|
83 |
|
|
<ENTRY key="mapping" ref="mappingPublication"/>
|
84 |
|
|
<ENTRY key="reuseMdRecords" ref="reusePublication"/>
|
85 |
|
|
<ENTRY key="hdfsPath" ref="hdfsPathPublication"/>
|
86 |
44367
|
claudio.at
|
</MAP>
|
87 |
|
|
</PARAM>
|
88 |
|
|
</PARAMETERS>
|
89 |
|
|
<ARCS>
|
90 |
45704
|
claudio.at
|
<ARC to="pmfResolved2hbase"/>
|
91 |
44367
|
claudio.at
|
</ARCS>
|
92 |
|
|
</NODE>
|
93 |
|
|
|
94 |
45704
|
claudio.at
|
<NODE name="pmfResolved2hbase" type="LaunchWorkflowTemplate">
|
95 |
|
|
<DESCRIPTION>PMF Resolved Publications to HBase</DESCRIPTION>
|
96 |
44367
|
claudio.at
|
<PARAMETERS>
|
97 |
|
|
<PARAM name="wfTemplateId" value="5b05a65a-4eeb-4862-bc55-b35c7ec3baf0_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ=="/>
|
98 |
|
|
<PARAM name="wfTemplateParams">
|
99 |
|
|
<MAP>
|
100 |
|
|
<ENTRY key="hbaseTable" ref="table"/>
|
101 |
|
|
<ENTRY key="cluster" ref="cluster"/>
|
102 |
|
|
<ENTRY key="mdFormat" value="PMF"/>
|
103 |
45704
|
claudio.at
|
<ENTRY key="interpretation" value="resolved"/>
|
104 |
44367
|
claudio.at
|
<ENTRY key="mapping" ref="mappingPublication"/>
|
105 |
45704
|
claudio.at
|
<ENTRY key="reuseMdRecords" ref="reuseResolvedPublication"/>
|
106 |
|
|
<ENTRY key="hdfsPath" ref="hdfsPathResolvedPublication"/>
|
107 |
44367
|
claudio.at
|
</MAP>
|
108 |
|
|
</PARAM>
|
109 |
|
|
</PARAMETERS>
|
110 |
|
|
<ARCS>
|
111 |
|
|
<ARC to="dmf2hbase"/>
|
112 |
|
|
</ARCS>
|
113 |
|
|
</NODE>
|
114 |
|
|
|
115 |
|
|
|
116 |
|
|
<NODE name="dmf2hbase" type="LaunchWorkflowTemplate">
|
117 |
45704
|
claudio.at
|
<DESCRIPTION>DMF Datasets to HBase</DESCRIPTION>
|
118 |
44367
|
claudio.at
|
<PARAMETERS>
|
119 |
|
|
<PARAM name="wfTemplateId" value="5b05a65a-4eeb-4862-bc55-b35c7ec3baf0_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ=="/>
|
120 |
|
|
<PARAM name="wfTemplateParams">
|
121 |
|
|
<MAP>
|
122 |
|
|
<ENTRY key="hbaseTable" ref="table"/>
|
123 |
|
|
<ENTRY key="cluster" ref="cluster"/>
|
124 |
|
|
<ENTRY key="mdFormat" value="DMF"/>
|
125 |
|
|
<ENTRY key="mapping" ref="mappingDataset"/>
|
126 |
|
|
<ENTRY key="reuseMdRecords" ref="reuseDataset"/>
|
127 |
|
|
<ENTRY key="hdfsPath" ref="hdfsPathDataset"/>
|
128 |
|
|
</MAP>
|
129 |
|
|
</PARAM>
|
130 |
|
|
</PARAMETERS>
|
131 |
|
|
<ARCS>
|
132 |
45704
|
claudio.at
|
<ARC to="dmfResolved2hbase"/>
|
133 |
44367
|
claudio.at
|
</ARCS>
|
134 |
|
|
</NODE>
|
135 |
|
|
|
136 |
45704
|
claudio.at
|
<NODE name="dmfResolved2hbase" type="LaunchWorkflowTemplate">
|
137 |
|
|
<DESCRIPTION>DMF Resolved Datasets to HBase</DESCRIPTION>
|
138 |
|
|
<PARAMETERS>
|
139 |
|
|
<PARAM name="wfTemplateId" value="5b05a65a-4eeb-4862-bc55-b35c7ec3baf0_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ=="/>
|
140 |
|
|
<PARAM name="wfTemplateParams">
|
141 |
|
|
<MAP>
|
142 |
|
|
<ENTRY key="hbaseTable" ref="table"/>
|
143 |
|
|
<ENTRY key="cluster" ref="cluster"/>
|
144 |
|
|
<ENTRY key="mdFormat" value="DMF"/>
|
145 |
|
|
<ENTRY key="interpretation" value="resolved"/>
|
146 |
|
|
<ENTRY key="mapping" ref="mappingDataset"/>
|
147 |
|
|
<ENTRY key="reuseMdRecords" ref="reuseResolvedDataset"/>
|
148 |
|
|
<ENTRY key="hdfsPath" ref="hdfsPathResolvedDataset"/>
|
149 |
|
|
</MAP>
|
150 |
|
|
</PARAM>
|
151 |
|
|
</PARAMETERS>
|
152 |
|
|
<ARCS>
|
153 |
|
|
<ARC to="scholix2hbase"/>
|
154 |
|
|
</ARCS>
|
155 |
|
|
</NODE>
|
156 |
|
|
|
157 |
|
|
<NODE name="scholix2hbase" type="LaunchWorkflowTemplate">
|
158 |
|
|
<DESCRIPTION>ODF Datasets to HBase</DESCRIPTION>
|
159 |
|
|
<PARAMETERS>
|
160 |
|
|
<PARAM name="wfTemplateId" value="5b05a65a-4eeb-4862-bc55-b35c7ec3baf0_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ=="/>
|
161 |
|
|
<PARAM name="wfTemplateParams">
|
162 |
|
|
<MAP>
|
163 |
|
|
<ENTRY key="hbaseTable" ref="table"/>
|
164 |
|
|
<ENTRY key="cluster" ref="cluster"/>
|
165 |
|
|
<ENTRY key="mdFormat" value="Scholix"/>
|
166 |
|
|
<ENTRY key="mapping" ref="mappingUnresolved"/>
|
167 |
|
|
<ENTRY key="reuseMdRecords" ref="reuseUnresolved"/>
|
168 |
|
|
<ENTRY key="hdfsPath" ref="hdfsPathUnresolved"/>
|
169 |
|
|
</MAP>
|
170 |
|
|
</PARAM>
|
171 |
|
|
</PARAMETERS>
|
172 |
|
|
<ARCS>
|
173 |
45917
|
sandro.lab
|
<ARC to="promoteActions"/>
|
174 |
45704
|
claudio.at
|
</ARCS>
|
175 |
|
|
</NODE>
|
176 |
|
|
|
177 |
|
|
<NODE name="promoteActions" type="PromoteActionsHDFS">
|
178 |
|
|
<DESCRIPTION>Promote Actions</DESCRIPTION>
|
179 |
|
|
<PARAMETERS>
|
180 |
|
|
<PARAM name="set" ref="actionSets"/>
|
181 |
|
|
<PARAM name="tableName" ref="table"/>
|
182 |
|
|
</PARAMETERS>
|
183 |
|
|
<ARCS>
|
184 |
|
|
<ARC to="closeMesh"/>
|
185 |
|
|
</ARCS>
|
186 |
|
|
</NODE>
|
187 |
|
|
|
188 |
|
|
<NODE name="closeMesh" type="LaunchWorkflowTemplate">
|
189 |
|
|
<DESCRIPTION>close mesh</DESCRIPTION>
|
190 |
|
|
<PARAMETERS>
|
191 |
45917
|
sandro.lab
|
<PARAM name="wfTemplateId"
|
192 |
|
|
value="70274106-375d-4135-9de1-536a606b327b_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ=="/>
|
193 |
45704
|
claudio.at
|
<PARAM name="wfTemplateParams">
|
194 |
|
|
<MAP>
|
195 |
|
|
<ENTRY key="hbaseTable" ref="table"/>
|
196 |
46058
|
sandro.lab
|
<ENTRY key="actionSets" ref="actionSets"/>
|
197 |
45704
|
claudio.at
|
<ENTRY key="cluster" ref="cluster"/>
|
198 |
|
|
<ENTRY key="workDir" ref="minDistWorkDir"/>
|
199 |
46058
|
sandro.lab
|
<ENTRY key="entitySequence" ref="entitysequence"/>
|
200 |
45704
|
claudio.at
|
</MAP>
|
201 |
|
|
</PARAM>
|
202 |
|
|
</PARAMETERS>
|
203 |
|
|
|
204 |
|
|
<ARCS>
|
205 |
46058
|
sandro.lab
|
<ARC to="fixRelations"/>
|
206 |
|
|
</ARCS>
|
207 |
|
|
</NODE>
|
208 |
|
|
|
209 |
|
|
|
210 |
|
|
<NODE name="fixRelations" type="LaunchWorkflowTemplate">
|
211 |
|
|
<DESCRIPTION>close mesh</DESCRIPTION>
|
212 |
|
|
<PARAMETERS>
|
213 |
|
|
<PARAM name="wfTemplateId"
|
214 |
|
|
value="caee3f24-1756-4a96-a25d-7c1b01c8d172_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ=="/>
|
215 |
|
|
<PARAM name="wfTemplateParams">
|
216 |
|
|
<MAP>
|
217 |
|
|
<ENTRY key="hbaseTable" ref="table"/>
|
218 |
|
|
<ENTRY key="actionSets" ref="actionSets"/>
|
219 |
|
|
<ENTRY key="cluster" ref="cluster"/>
|
220 |
|
|
<ENTRY key="workDir" ref="minDistWorkDir"/>
|
221 |
|
|
<ENTRY key="entitySequence" ref="entitysequence"/>
|
222 |
|
|
</MAP>
|
223 |
|
|
</PARAM>
|
224 |
|
|
</PARAMETERS>
|
225 |
|
|
|
226 |
|
|
<ARCS>
|
227 |
45897
|
claudio.at
|
<ARC to="updateIndex"/>
|
228 |
45704
|
claudio.at
|
</ARCS>
|
229 |
|
|
</NODE>
|
230 |
|
|
|
231 |
46058
|
sandro.lab
|
|
232 |
45897
|
claudio.at
|
<NODE name="updateIndex" type="LaunchWorkflowTemplate">
|
233 |
|
|
<DESCRIPTION>update provision</DESCRIPTION>
|
234 |
44367
|
claudio.at
|
<PARAMETERS>
|
235 |
45897
|
claudio.at
|
<PARAM name="wfTemplateId" value="bf449f4c-2da5-466f-aabc-e2269980e6b2_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ=="/>
|
236 |
44367
|
claudio.at
|
<PARAM name="wfTemplateParams">
|
237 |
|
|
<MAP>
|
238 |
|
|
<ENTRY key="hbaseTable" ref="table"/>
|
239 |
|
|
<ENTRY key="cluster" ref="cluster"/>
|
240 |
|
|
</MAP>
|
241 |
|
|
</PARAM>
|
242 |
|
|
</PARAMETERS>
|
243 |
|
|
<ARCS>
|
244 |
45897
|
claudio.at
|
<ARC to="success"/>
|
245 |
44367
|
claudio.at
|
</ARCS>
|
246 |
|
|
</NODE>
|
247 |
|
|
|
248 |
|
|
|
249 |
|
|
</WORKFLOW>
|
250 |
|
|
|
251 |
|
|
</CONFIGURATION>
|
252 |
|
|
|
253 |
|
|
<NOTIFICATIONS/>
|
254 |
|
|
|
255 |
|
|
<SCHEDULING enabled="false">
|
256 |
|
|
<CRON>29 5 22 ? * *</CRON>
|
257 |
|
|
<MININTERVAL>10080</MININTERVAL>
|
258 |
|
|
</SCHEDULING>
|
259 |
|
|
|
260 |
|
|
<STATUS/>
|
261 |
|
|
</BODY>
|
262 |
|
|
</RESOURCE_PROFILE>
|