1 |
29365
|
eri.katsar
|
<workflow-app name="test-core_examples_javamapreduce_stats" xmlns="uri:oozie:workflow:0.4">
|
2 |
|
|
<!-- map reduce job that exports hbase data and prepares them for import to the relation database used for statistics generation -->
|
3 |
29327
|
eri.katsar
|
|
4 |
28535
|
eri.katsar
|
<global>
|
5 |
|
|
<job-tracker>${jobTracker}</job-tracker>
|
6 |
|
|
<name-node>${nameNode}</name-node>
|
7 |
|
|
<configuration>
|
8 |
|
|
<property>
|
9 |
|
|
<name>mapred.job.queue.name</name>
|
10 |
|
|
<value>${queueName}</value>
|
11 |
|
|
</property>
|
12 |
29238
|
eri.katsar
|
<property>
|
13 |
|
|
<name>oozie.sqoop.log.level</name>
|
14 |
|
|
<value>DEBUG</value>
|
15 |
|
|
</property>
|
16 |
28523
|
eri.katsar
|
</configuration>
|
17 |
29238
|
eri.katsar
|
</global>
|
18 |
29521
|
eri.katsar
|
<start to='get-scanner' />
|
19 |
28512
|
eri.katsar
|
<action name='get-scanner'>
|
20 |
|
|
<java>
|
21 |
29327
|
eri.katsar
|
<main-class>eu.dnetlib.iis.core.workflows.stats.HbaseScannerGenerator
|
22 |
|
|
</main-class>
|
23 |
29238
|
eri.katsar
|
<!-- column families: -->
|
24 |
29500
|
eri.katsar
|
<arg> -f datasource
|
25 |
29502
|
eri.katsar
|
<!-- , -->
|
26 |
29460
|
eri.katsar
|
<!-- datasourceOrganization_provision_provides , -->
|
27 |
|
|
<!-- organization, -->
|
28 |
|
|
<!-- project, projectOrganization_participation_hasParticipant -->
|
29 |
29442
|
eri.katsar
|
<!-- result,resultProject_outcome_produces, -->
|
30 |
|
|
<!-- personResult_authorship_hasAuthor,resultResult_publicationDataset_isRelatedTo -->
|
31 |
29327
|
eri.katsar
|
</arg>
|
32 |
|
|
|
33 |
|
|
<capture-output />
|
34 |
28512
|
eri.katsar
|
</java>
|
35 |
|
|
<ok to="mr_export" />
|
36 |
|
|
<error to="fail" />
|
37 |
|
|
</action>
|
38 |
|
|
<action name="mr_export">
|
39 |
|
|
<map-reduce>
|
40 |
29327
|
eri.katsar
|
|
41 |
28514
|
eri.katsar
|
<prepare>
|
42 |
28565
|
eri.katsar
|
<delete path="${nameNode}${Stats_output_Path}" />
|
43 |
29327
|
eri.katsar
|
|
44 |
28514
|
eri.katsar
|
</prepare>
|
45 |
|
|
<configuration>
|
46 |
28512
|
eri.katsar
|
<property>
|
47 |
|
|
<name>hbase.mapreduce.scan</name>
|
48 |
|
|
<value>${wf:actionData('get-scanner')['scan']}</value>
|
49 |
|
|
</property>
|
50 |
28514
|
eri.katsar
|
<property>
|
51 |
|
|
<name>hbase.rootdir</name>
|
52 |
|
|
<value>hdfs://nmis-hadoop-cluster/hbase</value>
|
53 |
|
|
</property>
|
54 |
29327
|
eri.katsar
|
|
55 |
28514
|
eri.katsar
|
<property>
|
56 |
|
|
<name>hbase.security.authentication</name>
|
57 |
29238
|
eri.katsar
|
<value>simple</value>
|
58 |
28514
|
eri.katsar
|
</property>
|
59 |
28523
|
eri.katsar
|
<!-- ZOOKEEPER -->
|
60 |
29327
|
eri.katsar
|
|
61 |
28514
|
eri.katsar
|
<property>
|
62 |
|
|
<name>hbase.zookeeper.quorum</name>
|
63 |
29238
|
eri.katsar
|
<value>
|
64 |
|
|
quorum1.t.hadoop.research-infrastructures.eu,quorum2.t.hadoop.research-infrastructures.eu,quorum3.t.hadoop.research-infrastructures.eu,quorum4.t.hadoop.research-infrastructures.eu,jobtracker.t.hadoop.research-infrastructures.eu
|
65 |
28514
|
eri.katsar
|
</value>
|
66 |
|
|
</property>
|
67 |
|
|
<property>
|
68 |
|
|
<name>zookeeper.znode.rootserver</name>
|
69 |
29238
|
eri.katsar
|
<value>root-region-server</value>
|
70 |
28514
|
eri.katsar
|
</property>
|
71 |
29327
|
eri.katsar
|
|
72 |
28514
|
eri.katsar
|
<property>
|
73 |
|
|
<name>hbase.zookeeper.property.clientPort</name>
|
74 |
|
|
<value>2182</value>
|
75 |
|
|
</property>
|
76 |
29327
|
eri.katsar
|
|
77 |
|
|
|
78 |
28519
|
eri.katsar
|
<!-- MR IO -->
|
79 |
29327
|
eri.katsar
|
|
80 |
|
|
|
81 |
28514
|
eri.katsar
|
<property>
|
82 |
|
|
<name>mapreduce.inputformat.class</name>
|
83 |
|
|
<value>org.apache.hadoop.hbase.mapreduce.TableInputFormat</value>
|
84 |
|
|
</property>
|
85 |
29327
|
eri.katsar
|
|
86 |
28514
|
eri.katsar
|
<property>
|
87 |
|
|
<name>mapred.mapoutput.key.class</name>
|
88 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
89 |
|
|
</property>
|
90 |
|
|
<property>
|
91 |
|
|
<name>mapred.mapoutput.value.class</name>
|
92 |
|
|
<value>org.apache.hadoop.hbase.io.ImmutableBytesWritable</value>
|
93 |
|
|
</property>
|
94 |
|
|
<property>
|
95 |
|
|
<name>mapred.output.key.class</name>
|
96 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
97 |
|
|
</property>
|
98 |
|
|
<property>
|
99 |
|
|
<name>mapred.output.value.class</name>
|
100 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat
|
101 |
|
|
</value>
|
102 |
28514
|
eri.katsar
|
</property>
|
103 |
29327
|
eri.katsar
|
|
104 |
28514
|
eri.katsar
|
<!-- ## This is required for new MapReduce API usage -->
|
105 |
|
|
<property>
|
106 |
|
|
<name>mapred.mapper.new-api</name>
|
107 |
|
|
<value>true</value>
|
108 |
|
|
</property>
|
109 |
|
|
<property>
|
110 |
|
|
<name>mapred.reducer.new-api</name>
|
111 |
|
|
<value>true</value>
|
112 |
|
|
</property>
|
113 |
29327
|
eri.katsar
|
|
114 |
28514
|
eri.katsar
|
<!-- # Job-specific options -->
|
115 |
|
|
<property>
|
116 |
|
|
<name>dfs.blocksize</name>
|
117 |
|
|
<value>32M</value>
|
118 |
|
|
</property>
|
119 |
|
|
<property>
|
120 |
|
|
<name>mapred.output.compress</name>
|
121 |
|
|
<value>false</value>
|
122 |
|
|
</property>
|
123 |
|
|
<property>
|
124 |
|
|
<name>mapred.reduce.tasks.speculative.execution</name>
|
125 |
|
|
<value>false</value>
|
126 |
|
|
</property>
|
127 |
|
|
<property>
|
128 |
|
|
<name>mapred.reduce.tasks.speculative.execution</name>
|
129 |
|
|
<value>false</value>
|
130 |
|
|
</property>
|
131 |
|
|
<property>
|
132 |
|
|
<name>mapreduce.map.speculative</name>
|
133 |
|
|
<value>false</value>
|
134 |
|
|
</property>
|
135 |
29327
|
eri.katsar
|
|
136 |
28514
|
eri.katsar
|
<!-- I/O FORMAT -->
|
137 |
29365
|
eri.katsar
|
<!-- IMPORTANT: sets default delimeter used by text output writer. Required to fix issue with traling tab added between id and value in multiple outputs -->
|
138 |
29327
|
eri.katsar
|
<property>
|
139 |
29238
|
eri.katsar
|
<name>mapred.textoutputformat.separator</name>
|
140 |
|
|
<value>${Stats_delim_Character}</value>
|
141 |
|
|
</property>
|
142 |
28514
|
eri.katsar
|
<!-- ## Names of all output ports -->
|
143 |
29327
|
eri.katsar
|
|
144 |
28514
|
eri.katsar
|
<property>
|
145 |
28543
|
eri.katsar
|
<name>mapreduce.multipleoutputs</name>
|
146 |
29327
|
eri.katsar
|
|
147 |
29442
|
eri.katsar
|
<value>${out1} ${out2} ${out3} ${out4} ${out5} ${out6} ${out7} ${out8} ${out9} ${out10} ${out11} ${out12} ${out13} ${out14} ${out15} ${out16} ${out17} ${out18} ${out19} ${out20}
|
148 |
29327
|
eri.katsar
|
</value>
|
149 |
|
|
|
150 |
28539
|
eri.katsar
|
</property>
|
151 |
29238
|
eri.katsar
|
<!-- datasource -->
|
152 |
28539
|
eri.katsar
|
<property>
|
153 |
29238
|
eri.katsar
|
<name>mapreduce.multipleoutputs.namedOutput.${out1}.key</name>
|
154 |
28550
|
eri.katsar
|
<value>org.apache.hadoop.io.Text</value>
|
155 |
28539
|
eri.katsar
|
</property>
|
156 |
|
|
<property>
|
157 |
29238
|
eri.katsar
|
<name>mapreduce.multipleoutputs.namedOutput.${out1}.value</name>
|
158 |
28539
|
eri.katsar
|
<value>org.apache.hadoop.io.Text</value>
|
159 |
|
|
</property>
|
160 |
|
|
<property>
|
161 |
29238
|
eri.katsar
|
<name>mapreduce.multipleoutputs.namedOutput.${out1}.format</name>
|
162 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
163 |
|
|
</value>
|
164 |
28514
|
eri.katsar
|
</property>
|
165 |
29238
|
eri.katsar
|
<!-- datasourceLanguage -->
|
166 |
28514
|
eri.katsar
|
<property>
|
167 |
29238
|
eri.katsar
|
<name>mapreduce.multipleoutputs.namedOutput.${out2}.key</name>
|
168 |
28550
|
eri.katsar
|
<value>org.apache.hadoop.io.Text</value>
|
169 |
28514
|
eri.katsar
|
</property>
|
170 |
|
|
<property>
|
171 |
29238
|
eri.katsar
|
<name>mapreduce.multipleoutputs.namedOutput.${out2}.value</name>
|
172 |
28539
|
eri.katsar
|
<value>org.apache.hadoop.io.Text</value>
|
173 |
28514
|
eri.katsar
|
</property>
|
174 |
28539
|
eri.katsar
|
<property>
|
175 |
29238
|
eri.katsar
|
<name>mapreduce.multipleoutputs.namedOutput.${out2}.format</name>
|
176 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
177 |
|
|
</value>
|
178 |
28539
|
eri.katsar
|
</property>
|
179 |
29327
|
eri.katsar
|
|
180 |
|
|
|
181 |
|
|
|
182 |
29238
|
eri.katsar
|
<!-- datasourceOrganization -->
|
183 |
|
|
<property>
|
184 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out3}.key</name>
|
185 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
186 |
|
|
</property>
|
187 |
|
|
<property>
|
188 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out3}.value</name>
|
189 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
190 |
|
|
</property>
|
191 |
|
|
<property>
|
192 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out3}.format</name>
|
193 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
194 |
|
|
</value>
|
195 |
29238
|
eri.katsar
|
</property>
|
196 |
29327
|
eri.katsar
|
|
197 |
|
|
<!-- datasourceTopic -->
|
198 |
29238
|
eri.katsar
|
<property>
|
199 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out4}.key</name>
|
200 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
201 |
|
|
</property>
|
202 |
|
|
<property>
|
203 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out4}.value</name>
|
204 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
205 |
|
|
</property>
|
206 |
|
|
<property>
|
207 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out4}.format</name>
|
208 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
209 |
|
|
</value>
|
210 |
29238
|
eri.katsar
|
</property>
|
211 |
29327
|
eri.katsar
|
|
212 |
|
|
<!-- resultDatasource -->
|
213 |
29238
|
eri.katsar
|
<property>
|
214 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out5}.key</name>
|
215 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
216 |
|
|
</property>
|
217 |
|
|
<property>
|
218 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out5}.value</name>
|
219 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
220 |
|
|
</property>
|
221 |
|
|
<property>
|
222 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out5}.format</name>
|
223 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
224 |
|
|
</value>
|
225 |
29238
|
eri.katsar
|
</property>
|
226 |
29327
|
eri.katsar
|
<!-- organization -->
|
227 |
29238
|
eri.katsar
|
<property>
|
228 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out6}.key</name>
|
229 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
230 |
|
|
</property>
|
231 |
|
|
<property>
|
232 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out6}.value</name>
|
233 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
234 |
|
|
</property>
|
235 |
|
|
<property>
|
236 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out6}.format</name>
|
237 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
238 |
|
|
</value>
|
239 |
29238
|
eri.katsar
|
</property>
|
240 |
29327
|
eri.katsar
|
|
241 |
|
|
<!-- projectOrganization -->
|
242 |
29238
|
eri.katsar
|
<property>
|
243 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out7}.key</name>
|
244 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
245 |
|
|
</property>
|
246 |
|
|
<property>
|
247 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out7}.value</name>
|
248 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
249 |
|
|
</property>
|
250 |
|
|
<property>
|
251 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out7}.format</name>
|
252 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
253 |
|
|
</value>
|
254 |
29238
|
eri.katsar
|
</property>
|
255 |
29327
|
eri.katsar
|
<!-- resultProject -->
|
256 |
29238
|
eri.katsar
|
<property>
|
257 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out8}.key</name>
|
258 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
259 |
|
|
</property>
|
260 |
|
|
<property>
|
261 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out8}.value</name>
|
262 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
263 |
|
|
</property>
|
264 |
|
|
<property>
|
265 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out8}.format</name>
|
266 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
267 |
|
|
</value>
|
268 |
29238
|
eri.katsar
|
</property>
|
269 |
29327
|
eri.katsar
|
|
270 |
|
|
<!-- project -->
|
271 |
29238
|
eri.katsar
|
<property>
|
272 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out9}.key</name>
|
273 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
274 |
|
|
</property>
|
275 |
|
|
<property>
|
276 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out9}.value</name>
|
277 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
278 |
|
|
</property>
|
279 |
|
|
<property>
|
280 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out9}.format</name>
|
281 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
282 |
|
|
</value>
|
283 |
29238
|
eri.katsar
|
</property>
|
284 |
29327
|
eri.katsar
|
|
285 |
|
|
<!-- resultConcept -->
|
286 |
29238
|
eri.katsar
|
<property>
|
287 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out10}.key</name>
|
288 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
289 |
|
|
</property>
|
290 |
|
|
<property>
|
291 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out10}.value</name>
|
292 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
293 |
|
|
</property>
|
294 |
|
|
<property>
|
295 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out10}.format</name>
|
296 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
297 |
|
|
</value>
|
298 |
29238
|
eri.katsar
|
</property>
|
299 |
29327
|
eri.katsar
|
|
300 |
29238
|
eri.katsar
|
<!-- resultClaim -->
|
301 |
|
|
<property>
|
302 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out11}.key</name>
|
303 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
304 |
|
|
</property>
|
305 |
|
|
<property>
|
306 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out11}.value</name>
|
307 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
308 |
|
|
</property>
|
309 |
|
|
<property>
|
310 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out11}.format</name>
|
311 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
312 |
|
|
</value>
|
313 |
29238
|
eri.katsar
|
</property>
|
314 |
29327
|
eri.katsar
|
|
315 |
29238
|
eri.katsar
|
<!-- resultClassification -->
|
316 |
|
|
<property>
|
317 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out12}.key</name>
|
318 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
319 |
|
|
</property>
|
320 |
|
|
<property>
|
321 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out12}.value</name>
|
322 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
323 |
|
|
</property>
|
324 |
|
|
<property>
|
325 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out12}.format</name>
|
326 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
327 |
|
|
</value>
|
328 |
29238
|
eri.katsar
|
</property>
|
329 |
29327
|
eri.katsar
|
|
330 |
29238
|
eri.katsar
|
<!-- resultLanguage -->
|
331 |
|
|
<property>
|
332 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out13}.key</name>
|
333 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
334 |
|
|
</property>
|
335 |
|
|
<property>
|
336 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out13}.value</name>
|
337 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
338 |
|
|
</property>
|
339 |
|
|
<property>
|
340 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out13}.format</name>
|
341 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
342 |
|
|
</value>
|
343 |
29238
|
eri.katsar
|
</property>
|
344 |
29327
|
eri.katsar
|
|
345 |
|
|
<!-- resultProject -->
|
346 |
29238
|
eri.katsar
|
<property>
|
347 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out14}.key</name>
|
348 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
349 |
|
|
</property>
|
350 |
|
|
<property>
|
351 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out14}.value</name>
|
352 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
353 |
|
|
</property>
|
354 |
|
|
<property>
|
355 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out14}.format</name>
|
356 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
357 |
|
|
</value>
|
358 |
29238
|
eri.katsar
|
</property>
|
359 |
29327
|
eri.katsar
|
<!-- resultResult -->
|
360 |
29238
|
eri.katsar
|
<property>
|
361 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out15}.key</name>
|
362 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
363 |
|
|
</property>
|
364 |
|
|
<property>
|
365 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out15}.value</name>
|
366 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
367 |
|
|
</property>
|
368 |
|
|
<property>
|
369 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out15}.format</name>
|
370 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
371 |
|
|
</value>
|
372 |
29238
|
eri.katsar
|
</property>
|
373 |
29327
|
eri.katsar
|
<!-- resultTopic -->
|
374 |
29238
|
eri.katsar
|
<property>
|
375 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out16}.key</name>
|
376 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
377 |
|
|
</property>
|
378 |
|
|
<property>
|
379 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out16}.value</name>
|
380 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
381 |
|
|
</property>
|
382 |
|
|
<property>
|
383 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out16}.format</name>
|
384 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
385 |
|
|
</value>
|
386 |
29238
|
eri.katsar
|
</property>
|
387 |
29327
|
eri.katsar
|
|
388 |
|
|
<!-- resultDatasource -->
|
389 |
29238
|
eri.katsar
|
<property>
|
390 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out17}.key</name>
|
391 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
392 |
|
|
</property>
|
393 |
|
|
<property>
|
394 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out17}.value</name>
|
395 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
396 |
|
|
</property>
|
397 |
|
|
<property>
|
398 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out17}.format</name>
|
399 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
400 |
|
|
</value>
|
401 |
29238
|
eri.katsar
|
</property>
|
402 |
29327
|
eri.katsar
|
|
403 |
|
|
|
404 |
|
|
|
405 |
|
|
<!-- result -->
|
406 |
29238
|
eri.katsar
|
<property>
|
407 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out18}.key</name>
|
408 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
409 |
|
|
</property>
|
410 |
|
|
<property>
|
411 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out18}.value</name>
|
412 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
413 |
|
|
</property>
|
414 |
|
|
<property>
|
415 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out18}.format</name>
|
416 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
417 |
|
|
</value>
|
418 |
29238
|
eri.katsar
|
</property>
|
419 |
29327
|
eri.katsar
|
|
420 |
|
|
<!-- claim -->
|
421 |
29238
|
eri.katsar
|
<property>
|
422 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out19}.key</name>
|
423 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
424 |
|
|
</property>
|
425 |
|
|
<property>
|
426 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out19}.value</name>
|
427 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
428 |
|
|
</property>
|
429 |
|
|
<property>
|
430 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out19}.format</name>
|
431 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
432 |
|
|
</value>
|
433 |
29238
|
eri.katsar
|
</property>
|
434 |
29327
|
eri.katsar
|
<!-- cncept -->
|
435 |
29238
|
eri.katsar
|
<property>
|
436 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out20}.key</name>
|
437 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
438 |
|
|
</property>
|
439 |
|
|
<property>
|
440 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out20}.value</name>
|
441 |
|
|
<value>org.apache.hadoop.io.Text</value>
|
442 |
|
|
</property>
|
443 |
|
|
<property>
|
444 |
|
|
<name>mapreduce.multipleoutputs.namedOutput.${out20}.format</name>
|
445 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
446 |
|
|
</value>
|
447 |
29238
|
eri.katsar
|
</property>
|
448 |
29327
|
eri.katsar
|
|
449 |
29442
|
eri.katsar
|
|
450 |
28514
|
eri.katsar
|
<!-- ## Classes of mapper and reducer -->
|
451 |
29327
|
eri.katsar
|
|
452 |
28514
|
eri.katsar
|
<property>
|
453 |
|
|
<name>mapreduce.map.class</name>
|
454 |
29327
|
eri.katsar
|
<value>eu.dnetlib.data.mapreduce.hbase.statsExport.StatsMapper
|
455 |
|
|
</value>
|
456 |
28514
|
eri.katsar
|
</property>
|
457 |
|
|
<property>
|
458 |
|
|
<name>mapreduce.reduce.class</name>
|
459 |
29327
|
eri.katsar
|
<value>eu.dnetlib.data.mapreduce.hbase.statsExport.StatsReducer
|
460 |
|
|
</value>
|
461 |
28514
|
eri.katsar
|
</property>
|
462 |
28535
|
eri.katsar
|
<property>
|
463 |
|
|
<name>io.serializations</name>
|
464 |
29327
|
eri.katsar
|
<value>org.apache.hadoop.io.serializer.WritableSerialization
|
465 |
|
|
</value>
|
466 |
28535
|
eri.katsar
|
</property>
|
467 |
28519
|
eri.katsar
|
<!-- ## Custom config -->
|
468 |
29327
|
eri.katsar
|
|
469 |
28514
|
eri.katsar
|
<!--delim character used to seperate fields in hdfs dump files <property> -->
|
470 |
28519
|
eri.katsar
|
<property>
|
471 |
29238
|
eri.katsar
|
<name>stats.delim</name>
|
472 |
28565
|
eri.katsar
|
<value>${Stats_delim_Character}</value>
|
473 |
28519
|
eri.katsar
|
</property>
|
474 |
28514
|
eri.katsar
|
<!--default string for Null String Values -->
|
475 |
|
|
<property>
|
476 |
|
|
<name>stats.nullString</name>
|
477 |
28565
|
eri.katsar
|
<value>${Stats_null_String_Field}</value>
|
478 |
28514
|
eri.katsar
|
</property>
|
479 |
|
|
<!--default string for Null Numeric Values -->
|
480 |
|
|
<property>
|
481 |
|
|
<name>stats.nullNum</name>
|
482 |
28565
|
eri.katsar
|
<value>${Stats_null_Numeric_Field}</value>
|
483 |
28514
|
eri.katsar
|
</property>
|
484 |
29365
|
eri.katsar
|
<property>
|
485 |
|
|
<name>stats.enclChar</name>
|
486 |
|
|
<value>${Stats_enclosing_Character}</value>
|
487 |
|
|
</property>
|
488 |
|
|
|
489 |
|
|
|
490 |
28514
|
eri.katsar
|
<!--source hbase table -->
|
491 |
29498
|
eri.katsar
|
<property>
|
492 |
28514
|
eri.katsar
|
<name>hbase.mapreduce.inputtable</name>
|
493 |
28565
|
eri.katsar
|
<value>${Stats_Hbase_Source_Table}</value>
|
494 |
28514
|
eri.katsar
|
</property>
|
495 |
29500
|
eri.katsar
|
<property>
|
496 |
29497
|
eri.katsar
|
<name>hbase.mapred.inputtable</name>
|
497 |
|
|
<value>${Stats_Hbase_Source_Table}</value>
|
498 |
|
|
</property>
|
499 |
29498
|
eri.katsar
|
|
500 |
29327
|
eri.katsar
|
<property>
|
501 |
|
|
<!-- mapping of protos entities to tables in the relDB -->
|
502 |
|
|
<name>stats.dbTablesMap</name>
|
503 |
|
|
<value>${Stats_db_table_map}</value>
|
504 |
29249
|
eri.katsar
|
</property>
|
505 |
29365
|
eri.katsar
|
|
506 |
|
|
<!-- This directory does not correspond to a data store. In fact, this directory only contains multiple data stores. It has to be set to the name of the workflow node. -->
|
507 |
28514
|
eri.katsar
|
<property>
|
508 |
|
|
<name>mapred.output.dir</name>
|
509 |
28565
|
eri.katsar
|
<value>${Stats_output_Path}</value>
|
510 |
28514
|
eri.katsar
|
</property>
|
511 |
29327
|
eri.katsar
|
<property>
|
512 |
|
|
<name>stats.indexConf</name>
|
513 |
|
|
<value>${Stats_indexConf}</value>
|
514 |
|
|
</property>
|
515 |
28514
|
eri.katsar
|
<!-- ## Workflow node parameters -->
|
516 |
|
|
<property>
|
517 |
|
|
<name>mapred.reduce.tasks</name>
|
518 |
29327
|
eri.katsar
|
<value>${numReducers}</value>
|
519 |
28514
|
eri.katsar
|
</property>
|
520 |
29510
|
eri.katsar
|
|
521 |
28514
|
eri.katsar
|
</configuration>
|
522 |
28512
|
eri.katsar
|
</map-reduce>
|
523 |
29442
|
eri.katsar
|
<ok to="exportContext" />
|
524 |
|
|
<error to="fail" />
|
525 |
|
|
</action>
|
526 |
|
|
|
527 |
|
|
|
528 |
|
|
<action name="exportContext">
|
529 |
|
|
<java>
|
530 |
|
|
|
531 |
|
|
<prepare>
|
532 |
|
|
</prepare>
|
533 |
|
|
<configuration>
|
534 |
|
|
<property>
|
535 |
|
|
<name>mapred.job.queue.name</name>
|
536 |
|
|
<value>${queueName}</value>
|
537 |
|
|
</property>
|
538 |
|
|
</configuration>
|
539 |
|
|
|
540 |
|
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
541 |
29505
|
eri.katsar
|
<arg>-SworkingDir=${workingDir}</arg>
|
542 |
29442
|
eri.katsar
|
<arg>eu.dnetlib.iis.core.workflows.stats.ExportContextWrapper</arg>
|
543 |
|
|
<arg>-PStats_output_Path=${Stats_output_Path}</arg>
|
544 |
|
|
<arg>-PStats_ContextResourceXML=${ContextResourceXML}</arg>
|
545 |
|
|
|
546 |
|
|
</java>
|
547 |
29412
|
eri.katsar
|
<ok to="prepareDatabase" />
|
548 |
28512
|
eri.katsar
|
<error to="fail" />
|
549 |
|
|
</action>
|
550 |
29327
|
eri.katsar
|
|
551 |
28543
|
eri.katsar
|
<action name="prepareDatabase">
|
552 |
28499
|
eri.katsar
|
<java>
|
553 |
29327
|
eri.katsar
|
|
554 |
28499
|
eri.katsar
|
<prepare>
|
555 |
|
|
</prepare>
|
556 |
|
|
<configuration>
|
557 |
|
|
<property>
|
558 |
|
|
<name>mapred.job.queue.name</name>
|
559 |
|
|
<value>${queueName}</value>
|
560 |
|
|
</property>
|
561 |
|
|
</configuration>
|
562 |
29327
|
eri.katsar
|
|
563 |
29238
|
eri.katsar
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
564 |
29505
|
eri.katsar
|
<arg>-SworkingDir=${workingDir}</arg>
|
565 |
28499
|
eri.katsar
|
<arg>eu.dnetlib.iis.core.workflows.stats.DBInitWrapper</arg>
|
566 |
29327
|
eri.katsar
|
|
567 |
28565
|
eri.katsar
|
<arg>-PStats_db_Url=${Stats_db_Url}</arg>
|
568 |
|
|
<arg>-PStats_db_User=${Stats_db_User}</arg>
|
569 |
|
|
<arg>-PStats_db_Pass=${Stats_db_Pass}</arg>
|
570 |
|
|
<arg>-PStats_db_Driver=${Stats_db_Driver}</arg>
|
571 |
29327
|
eri.katsar
|
|
572 |
28499
|
eri.katsar
|
</java>
|
573 |
28611
|
eri.katsar
|
<ok to="sqoopImport" />
|
574 |
28499
|
eri.katsar
|
<error to="fail" />
|
575 |
|
|
</action>
|
576 |
29327
|
eri.katsar
|
|
577 |
28499
|
eri.katsar
|
<action name="sqoopImport">
|
578 |
|
|
<java>
|
579 |
28571
|
eri.katsar
|
<prepare>
|
580 |
|
|
</prepare>
|
581 |
28499
|
eri.katsar
|
<configuration>
|
582 |
|
|
<property>
|
583 |
|
|
<name>mapred.job.queue.name</name>
|
584 |
|
|
<value>${queueName}</value>
|
585 |
|
|
</property>
|
586 |
29327
|
eri.katsar
|
|
587 |
29238
|
eri.katsar
|
<property>
|
588 |
|
|
<name>oozie.sqoop.log.level</name>
|
589 |
|
|
<value>DEBUG</value>
|
590 |
|
|
</property>
|
591 |
29327
|
eri.katsar
|
|
592 |
28499
|
eri.katsar
|
</configuration>
|
593 |
29327
|
eri.katsar
|
|
594 |
29238
|
eri.katsar
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
595 |
29510
|
eri.katsar
|
<arg>-SworkingDir=${workingDir}</arg>
|
596 |
|
|
<arg>eu.dnetlib.iis.core.workflows.stats.SqoopWrapper</arg>
|
597 |
|
|
<arg>-PStats_db_Url=${Stats_db_Url}</arg>
|
598 |
28565
|
eri.katsar
|
<arg>-PStats_db_User=${Stats_db_User}</arg>
|
599 |
|
|
<arg>-PStats_db_Pass=${Stats_db_Pass}</arg>
|
600 |
29327
|
eri.katsar
|
|
601 |
28605
|
eri.katsar
|
<arg>-PStats_output_Path=${Stats_output_Path}</arg>
|
602 |
29327
|
eri.katsar
|
<arg>-PStats_sqoop_RecsPerStatement=${Stats_sqoop_RecsPerStatement}
|
603 |
|
|
</arg>
|
604 |
28605
|
eri.katsar
|
<arg>-PStats_sqoop_ReducersCount=${Stats_sqoop_ReducersCount}</arg>
|
605 |
29327
|
eri.katsar
|
<arg>-PStats_sqoop_StatementPerTrans=${Stats_sqoop_StatementPerTrans}
|
606 |
|
|
</arg>
|
607 |
29238
|
eri.katsar
|
<arg>-PStats_delim_Character=${Stats_delim_Character}</arg>
|
608 |
29327
|
eri.katsar
|
<arg>-PStats_db_table_map=${Stats_db_table_map}</arg>
|
609 |
29365
|
eri.katsar
|
<arg>-PStats_enclosing_Character=${Stats_enclosing_Character} </arg>
|
610 |
|
|
|
611 |
|
|
</java>
|
612 |
29412
|
eri.katsar
|
<ok to="finalizeDatabase" />
|
613 |
28499
|
eri.katsar
|
<error to="fail" />
|
614 |
|
|
</action>
|
615 |
29327
|
eri.katsar
|
|
616 |
28543
|
eri.katsar
|
<action name="finalizeDatabase">
|
617 |
28499
|
eri.katsar
|
<java>
|
618 |
29327
|
eri.katsar
|
|
619 |
28499
|
eri.katsar
|
<prepare>
|
620 |
|
|
</prepare>
|
621 |
|
|
<configuration>
|
622 |
|
|
<property>
|
623 |
|
|
<name>mapred.job.queue.name</name>
|
624 |
|
|
<value>${queueName}</value>
|
625 |
|
|
</property>
|
626 |
|
|
</configuration>
|
627 |
29327
|
eri.katsar
|
|
628 |
29238
|
eri.katsar
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
629 |
29505
|
eri.katsar
|
<arg>-SworkingDir=${workingDir}</arg>
|
630 |
29238
|
eri.katsar
|
<arg>eu.dnetlib.iis.core.workflows.stats.DBFinalizeWrapper</arg>
|
631 |
28565
|
eri.katsar
|
<arg>-PStats_db_Url=${Stats_db_Url}</arg>
|
632 |
|
|
<arg>-PStats_db_User=${Stats_db_User}</arg>
|
633 |
|
|
<arg>-PStats_db_Pass=${Stats_db_Pass}</arg>
|
634 |
29327
|
eri.katsar
|
<arg>-PStats_db_Driver=${Stats_db_Driver}</arg>
|
635 |
29442
|
eri.katsar
|
|
636 |
29327
|
eri.katsar
|
|
637 |
28499
|
eri.katsar
|
</java>
|
638 |
|
|
<ok to="end" />
|
639 |
|
|
<error to="fail" />
|
640 |
|
|
</action>
|
641 |
29442
|
eri.katsar
|
<kill name="fail">
|
642 |
28499
|
eri.katsar
|
<message>
|
643 |
29327
|
eri.katsar
|
Unfortunately, the process failed -- error message:
|
644 |
|
|
[${wf:errorMessage(wf:lastErrorNode())}]
|
645 |
28499
|
eri.katsar
|
</message>
|
646 |
|
|
</kill>
|
647 |
|
|
<end name="end" />
|
648 |
29412
|
eri.katsar
|
</workflow-app>
|