|
1
|
<workflow-app name="test-core_examples_javamapreduce_stats"
|
|
2
|
xmlns="uri:oozie:workflow:0.4">
|
|
3
|
<!-- map reduce job that exports hbase data and prepares them for import
|
|
4
|
to the relation database used for statistics generation -->
|
|
5
|
|
|
6
|
<global>
|
|
7
|
<job-tracker>${jobTracker}</job-tracker>
|
|
8
|
<name-node>${nameNode}</name-node>
|
|
9
|
<configuration>
|
|
10
|
<property>
|
|
11
|
<name>mapred.job.queue.name</name>
|
|
12
|
<value>${queueName}</value>
|
|
13
|
</property>
|
|
14
|
<property>
|
|
15
|
<name>oozie.sqoop.log.level</name>
|
|
16
|
<value>DEBUG</value>
|
|
17
|
</property>
|
|
18
|
</configuration>
|
|
19
|
</global>
|
|
20
|
<start to='get-scanner' />
|
|
21
|
<action name='get-scanner'>
|
|
22
|
<java>
|
|
23
|
<main-class>eu.dnetlib.iis.core.workflows.stats.HbaseScannerGenerator
|
|
24
|
</main-class>
|
|
25
|
<!-- column families: -->
|
|
26
|
|
|
27
|
<arg>
|
|
28
|
-f datasource
|
|
29
|
<!-- , datasourceOrganization_provision_provides ,organization, -->
|
|
30
|
<!-- projectOrganization_participation_isParticipant, -->
|
|
31
|
<!-- project -->
|
|
32
|
<!-- ,projectOrganization_participation_hasParticipant -->
|
|
33
|
<!-- , -->
|
|
34
|
<!-- result -->
|
|
35
|
<!-- , resultProject_outcome_produces, -->
|
|
36
|
<!-- personResult_authorship_hasAuthor,resultResult_publicationDataset_isRelatedTo -->
|
|
37
|
</arg>
|
|
38
|
|
|
39
|
<capture-output />
|
|
40
|
</java>
|
|
41
|
<ok to="mr_export" />
|
|
42
|
<error to="fail" />
|
|
43
|
</action>
|
|
44
|
<action name="mr_export">
|
|
45
|
<map-reduce>
|
|
46
|
|
|
47
|
<prepare>
|
|
48
|
<delete path="${nameNode}${Stats_output_Path}" />
|
|
49
|
|
|
50
|
</prepare>
|
|
51
|
<configuration>
|
|
52
|
<property>
|
|
53
|
<name>hbase.mapreduce.scan</name>
|
|
54
|
<value>${wf:actionData('get-scanner')['scan']}</value>
|
|
55
|
</property>
|
|
56
|
<property>
|
|
57
|
<name>hbase.rootdir</name>
|
|
58
|
<value>hdfs://nmis-hadoop-cluster/hbase</value>
|
|
59
|
</property>
|
|
60
|
|
|
61
|
<property>
|
|
62
|
<name>hbase.security.authentication</name>
|
|
63
|
<value>simple</value>
|
|
64
|
</property>
|
|
65
|
<!-- ZOOKEEPER -->
|
|
66
|
|
|
67
|
<property>
|
|
68
|
<name>hbase.zookeeper.quorum</name>
|
|
69
|
<value>
|
|
70
|
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
|
|
71
|
</value>
|
|
72
|
</property>
|
|
73
|
<property>
|
|
74
|
<name>zookeeper.znode.rootserver</name>
|
|
75
|
<value>root-region-server</value>
|
|
76
|
</property>
|
|
77
|
|
|
78
|
<property>
|
|
79
|
<name>hbase.zookeeper.property.clientPort</name>
|
|
80
|
<value>2182</value>
|
|
81
|
</property>
|
|
82
|
|
|
83
|
|
|
84
|
<!-- MR IO -->
|
|
85
|
|
|
86
|
|
|
87
|
<property>
|
|
88
|
<name>mapreduce.inputformat.class</name>
|
|
89
|
<value>org.apache.hadoop.hbase.mapreduce.TableInputFormat</value>
|
|
90
|
</property>
|
|
91
|
|
|
92
|
<property>
|
|
93
|
<name>mapred.mapoutput.key.class</name>
|
|
94
|
<value>org.apache.hadoop.io.Text</value>
|
|
95
|
</property>
|
|
96
|
<property>
|
|
97
|
<name>mapred.mapoutput.value.class</name>
|
|
98
|
<value>org.apache.hadoop.hbase.io.ImmutableBytesWritable</value>
|
|
99
|
</property>
|
|
100
|
<property>
|
|
101
|
<name>mapred.output.key.class</name>
|
|
102
|
<value>org.apache.hadoop.io.Text</value>
|
|
103
|
</property>
|
|
104
|
<property>
|
|
105
|
<name>mapred.output.value.class</name>
|
|
106
|
<value>org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat
|
|
107
|
</value>
|
|
108
|
</property>
|
|
109
|
|
|
110
|
<!-- ## This is required for new MapReduce API usage -->
|
|
111
|
<property>
|
|
112
|
<name>mapred.mapper.new-api</name>
|
|
113
|
<value>true</value>
|
|
114
|
</property>
|
|
115
|
<property>
|
|
116
|
<name>mapred.reducer.new-api</name>
|
|
117
|
<value>true</value>
|
|
118
|
</property>
|
|
119
|
|
|
120
|
<!-- # Job-specific options -->
|
|
121
|
<property>
|
|
122
|
<name>dfs.blocksize</name>
|
|
123
|
<value>32M</value>
|
|
124
|
</property>
|
|
125
|
<property>
|
|
126
|
<name>mapred.output.compress</name>
|
|
127
|
<value>false</value>
|
|
128
|
</property>
|
|
129
|
<property>
|
|
130
|
<name>mapred.reduce.tasks.speculative.execution</name>
|
|
131
|
<value>false</value>
|
|
132
|
</property>
|
|
133
|
<property>
|
|
134
|
<name>mapred.reduce.tasks.speculative.execution</name>
|
|
135
|
<value>false</value>
|
|
136
|
</property>
|
|
137
|
<property>
|
|
138
|
<name>mapreduce.map.speculative</name>
|
|
139
|
<value>false</value>
|
|
140
|
</property>
|
|
141
|
|
|
142
|
<!-- I/O FORMAT -->
|
|
143
|
<!-- IMPORTANT: sets default delimeter used by text output writer. Required
|
|
144
|
to fix issue with traling tab added between id and value in multiple outputs -->
|
|
145
|
<property>
|
|
146
|
<name>mapred.textoutputformat.separator</name>
|
|
147
|
<value>${Stats_delim_Character}</value>
|
|
148
|
</property>
|
|
149
|
<!-- ## Names of all output ports -->
|
|
150
|
|
|
151
|
<property>
|
|
152
|
<name>mapreduce.multipleoutputs</name>
|
|
153
|
|
|
154
|
<value>${out1} ${out2} ${out3} ${out4} ${out5} ${out6} ${out7}
|
|
155
|
${out8} ${out9} ${out10} ${out11} ${out12} ${out13} ${out14}
|
|
156
|
${out15} ${out16} ${out17} ${out18} ${out19} ${out20} ${out21}
|
|
157
|
</value>
|
|
158
|
|
|
159
|
</property>
|
|
160
|
<!-- datasource -->
|
|
161
|
<property>
|
|
162
|
<name>mapreduce.multipleoutputs.namedOutput.${out1}.key</name>
|
|
163
|
<value>org.apache.hadoop.io.Text</value>
|
|
164
|
</property>
|
|
165
|
<property>
|
|
166
|
<name>mapreduce.multipleoutputs.namedOutput.${out1}.value</name>
|
|
167
|
<value>org.apache.hadoop.io.Text</value>
|
|
168
|
</property>
|
|
169
|
<property>
|
|
170
|
<name>mapreduce.multipleoutputs.namedOutput.${out1}.format</name>
|
|
171
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
172
|
</value>
|
|
173
|
</property>
|
|
174
|
<!-- datasourceLanguage -->
|
|
175
|
<property>
|
|
176
|
<name>mapreduce.multipleoutputs.namedOutput.${out2}.key</name>
|
|
177
|
<value>org.apache.hadoop.io.Text</value>
|
|
178
|
</property>
|
|
179
|
<property>
|
|
180
|
<name>mapreduce.multipleoutputs.namedOutput.${out2}.value</name>
|
|
181
|
<value>org.apache.hadoop.io.Text</value>
|
|
182
|
</property>
|
|
183
|
<property>
|
|
184
|
<name>mapreduce.multipleoutputs.namedOutput.${out2}.format</name>
|
|
185
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
186
|
</value>
|
|
187
|
</property>
|
|
188
|
|
|
189
|
|
|
190
|
|
|
191
|
<!-- datasourceOrganization -->
|
|
192
|
<property>
|
|
193
|
<name>mapreduce.multipleoutputs.namedOutput.${out3}.key</name>
|
|
194
|
<value>org.apache.hadoop.io.Text</value>
|
|
195
|
</property>
|
|
196
|
<property>
|
|
197
|
<name>mapreduce.multipleoutputs.namedOutput.${out3}.value</name>
|
|
198
|
<value>org.apache.hadoop.io.Text</value>
|
|
199
|
</property>
|
|
200
|
<property>
|
|
201
|
<name>mapreduce.multipleoutputs.namedOutput.${out3}.format</name>
|
|
202
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
203
|
</value>
|
|
204
|
</property>
|
|
205
|
|
|
206
|
<!-- datasourceTopic -->
|
|
207
|
<property>
|
|
208
|
<name>mapreduce.multipleoutputs.namedOutput.${out4}.key</name>
|
|
209
|
<value>org.apache.hadoop.io.Text</value>
|
|
210
|
</property>
|
|
211
|
<property>
|
|
212
|
<name>mapreduce.multipleoutputs.namedOutput.${out4}.value</name>
|
|
213
|
<value>org.apache.hadoop.io.Text</value>
|
|
214
|
</property>
|
|
215
|
<property>
|
|
216
|
<name>mapreduce.multipleoutputs.namedOutput.${out4}.format</name>
|
|
217
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
218
|
</value>
|
|
219
|
</property>
|
|
220
|
|
|
221
|
<!-- resultDatasource -->
|
|
222
|
<property>
|
|
223
|
<name>mapreduce.multipleoutputs.namedOutput.${out5}.key</name>
|
|
224
|
<value>org.apache.hadoop.io.Text</value>
|
|
225
|
</property>
|
|
226
|
<property>
|
|
227
|
<name>mapreduce.multipleoutputs.namedOutput.${out5}.value</name>
|
|
228
|
<value>org.apache.hadoop.io.Text</value>
|
|
229
|
</property>
|
|
230
|
<property>
|
|
231
|
<name>mapreduce.multipleoutputs.namedOutput.${out5}.format</name>
|
|
232
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
233
|
</value>
|
|
234
|
</property>
|
|
235
|
<!-- organization -->
|
|
236
|
<property>
|
|
237
|
<name>mapreduce.multipleoutputs.namedOutput.${out6}.key</name>
|
|
238
|
<value>org.apache.hadoop.io.Text</value>
|
|
239
|
</property>
|
|
240
|
<property>
|
|
241
|
<name>mapreduce.multipleoutputs.namedOutput.${out6}.value</name>
|
|
242
|
<value>org.apache.hadoop.io.Text</value>
|
|
243
|
</property>
|
|
244
|
<property>
|
|
245
|
<name>mapreduce.multipleoutputs.namedOutput.${out6}.format</name>
|
|
246
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
247
|
</value>
|
|
248
|
</property>
|
|
249
|
|
|
250
|
<!-- projectOrganization -->
|
|
251
|
<property>
|
|
252
|
<name>mapreduce.multipleoutputs.namedOutput.${out7}.key</name>
|
|
253
|
<value>org.apache.hadoop.io.Text</value>
|
|
254
|
</property>
|
|
255
|
<property>
|
|
256
|
<name>mapreduce.multipleoutputs.namedOutput.${out7}.value</name>
|
|
257
|
<value>org.apache.hadoop.io.Text</value>
|
|
258
|
</property>
|
|
259
|
<property>
|
|
260
|
<name>mapreduce.multipleoutputs.namedOutput.${out7}.format</name>
|
|
261
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
262
|
</value>
|
|
263
|
</property>
|
|
264
|
<!-- resultProject -->
|
|
265
|
<property>
|
|
266
|
<name>mapreduce.multipleoutputs.namedOutput.${out8}.key</name>
|
|
267
|
<value>org.apache.hadoop.io.Text</value>
|
|
268
|
</property>
|
|
269
|
<property>
|
|
270
|
<name>mapreduce.multipleoutputs.namedOutput.${out8}.value</name>
|
|
271
|
<value>org.apache.hadoop.io.Text</value>
|
|
272
|
</property>
|
|
273
|
<property>
|
|
274
|
<name>mapreduce.multipleoutputs.namedOutput.${out8}.format</name>
|
|
275
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
276
|
</value>
|
|
277
|
</property>
|
|
278
|
|
|
279
|
<!-- project -->
|
|
280
|
<property>
|
|
281
|
<name>mapreduce.multipleoutputs.namedOutput.${out9}.key</name>
|
|
282
|
<value>org.apache.hadoop.io.Text</value>
|
|
283
|
</property>
|
|
284
|
<property>
|
|
285
|
<name>mapreduce.multipleoutputs.namedOutput.${out9}.value</name>
|
|
286
|
<value>org.apache.hadoop.io.Text</value>
|
|
287
|
</property>
|
|
288
|
<property>
|
|
289
|
<name>mapreduce.multipleoutputs.namedOutput.${out9}.format</name>
|
|
290
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
291
|
</value>
|
|
292
|
</property>
|
|
293
|
|
|
294
|
<!-- resultConcept -->
|
|
295
|
<property>
|
|
296
|
<name>mapreduce.multipleoutputs.namedOutput.${out10}.key</name>
|
|
297
|
<value>org.apache.hadoop.io.Text</value>
|
|
298
|
</property>
|
|
299
|
<property>
|
|
300
|
<name>mapreduce.multipleoutputs.namedOutput.${out10}.value</name>
|
|
301
|
<value>org.apache.hadoop.io.Text</value>
|
|
302
|
</property>
|
|
303
|
<property>
|
|
304
|
<name>mapreduce.multipleoutputs.namedOutput.${out10}.format</name>
|
|
305
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
306
|
</value>
|
|
307
|
</property>
|
|
308
|
|
|
309
|
<!-- resultClaim -->
|
|
310
|
<property>
|
|
311
|
<name>mapreduce.multipleoutputs.namedOutput.${out11}.key</name>
|
|
312
|
<value>org.apache.hadoop.io.Text</value>
|
|
313
|
</property>
|
|
314
|
<property>
|
|
315
|
<name>mapreduce.multipleoutputs.namedOutput.${out11}.value</name>
|
|
316
|
<value>org.apache.hadoop.io.Text</value>
|
|
317
|
</property>
|
|
318
|
<property>
|
|
319
|
<name>mapreduce.multipleoutputs.namedOutput.${out11}.format</name>
|
|
320
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
321
|
</value>
|
|
322
|
</property>
|
|
323
|
|
|
324
|
<!-- resultClassification -->
|
|
325
|
<property>
|
|
326
|
<name>mapreduce.multipleoutputs.namedOutput.${out12}.key</name>
|
|
327
|
<value>org.apache.hadoop.io.Text</value>
|
|
328
|
</property>
|
|
329
|
<property>
|
|
330
|
<name>mapreduce.multipleoutputs.namedOutput.${out12}.value</name>
|
|
331
|
<value>org.apache.hadoop.io.Text</value>
|
|
332
|
</property>
|
|
333
|
<property>
|
|
334
|
<name>mapreduce.multipleoutputs.namedOutput.${out12}.format</name>
|
|
335
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
336
|
</value>
|
|
337
|
</property>
|
|
338
|
|
|
339
|
<!-- resultLanguage -->
|
|
340
|
<property>
|
|
341
|
<name>mapreduce.multipleoutputs.namedOutput.${out13}.key</name>
|
|
342
|
<value>org.apache.hadoop.io.Text</value>
|
|
343
|
</property>
|
|
344
|
<property>
|
|
345
|
<name>mapreduce.multipleoutputs.namedOutput.${out13}.value</name>
|
|
346
|
<value>org.apache.hadoop.io.Text</value>
|
|
347
|
</property>
|
|
348
|
<property>
|
|
349
|
<name>mapreduce.multipleoutputs.namedOutput.${out13}.format</name>
|
|
350
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
351
|
</value>
|
|
352
|
</property>
|
|
353
|
|
|
354
|
<!-- resultProject -->
|
|
355
|
<property>
|
|
356
|
<name>mapreduce.multipleoutputs.namedOutput.${out14}.key</name>
|
|
357
|
<value>org.apache.hadoop.io.Text</value>
|
|
358
|
</property>
|
|
359
|
<property>
|
|
360
|
<name>mapreduce.multipleoutputs.namedOutput.${out14}.value</name>
|
|
361
|
<value>org.apache.hadoop.io.Text</value>
|
|
362
|
</property>
|
|
363
|
<property>
|
|
364
|
<name>mapreduce.multipleoutputs.namedOutput.${out14}.format</name>
|
|
365
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
366
|
</value>
|
|
367
|
</property>
|
|
368
|
<!-- resultResult -->
|
|
369
|
<property>
|
|
370
|
<name>mapreduce.multipleoutputs.namedOutput.${out15}.key</name>
|
|
371
|
<value>org.apache.hadoop.io.Text</value>
|
|
372
|
</property>
|
|
373
|
<property>
|
|
374
|
<name>mapreduce.multipleoutputs.namedOutput.${out15}.value</name>
|
|
375
|
<value>org.apache.hadoop.io.Text</value>
|
|
376
|
</property>
|
|
377
|
<property>
|
|
378
|
<name>mapreduce.multipleoutputs.namedOutput.${out15}.format</name>
|
|
379
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
380
|
</value>
|
|
381
|
</property>
|
|
382
|
<!-- resultTopic -->
|
|
383
|
<property>
|
|
384
|
<name>mapreduce.multipleoutputs.namedOutput.${out16}.key</name>
|
|
385
|
<value>org.apache.hadoop.io.Text</value>
|
|
386
|
</property>
|
|
387
|
<property>
|
|
388
|
<name>mapreduce.multipleoutputs.namedOutput.${out16}.value</name>
|
|
389
|
<value>org.apache.hadoop.io.Text</value>
|
|
390
|
</property>
|
|
391
|
<property>
|
|
392
|
<name>mapreduce.multipleoutputs.namedOutput.${out16}.format</name>
|
|
393
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
394
|
</value>
|
|
395
|
</property>
|
|
396
|
|
|
397
|
<!-- resultDatasource -->
|
|
398
|
<property>
|
|
399
|
<name>mapreduce.multipleoutputs.namedOutput.${out17}.key</name>
|
|
400
|
<value>org.apache.hadoop.io.Text</value>
|
|
401
|
</property>
|
|
402
|
<property>
|
|
403
|
<name>mapreduce.multipleoutputs.namedOutput.${out17}.value</name>
|
|
404
|
<value>org.apache.hadoop.io.Text</value>
|
|
405
|
</property>
|
|
406
|
<property>
|
|
407
|
<name>mapreduce.multipleoutputs.namedOutput.${out17}.format</name>
|
|
408
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
409
|
</value>
|
|
410
|
</property>
|
|
411
|
|
|
412
|
|
|
413
|
|
|
414
|
<!-- result -->
|
|
415
|
<property>
|
|
416
|
<name>mapreduce.multipleoutputs.namedOutput.${out18}.key</name>
|
|
417
|
<value>org.apache.hadoop.io.Text</value>
|
|
418
|
</property>
|
|
419
|
<property>
|
|
420
|
<name>mapreduce.multipleoutputs.namedOutput.${out18}.value</name>
|
|
421
|
<value>org.apache.hadoop.io.Text</value>
|
|
422
|
</property>
|
|
423
|
<property>
|
|
424
|
<name>mapreduce.multipleoutputs.namedOutput.${out18}.format</name>
|
|
425
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
426
|
</value>
|
|
427
|
</property>
|
|
428
|
|
|
429
|
<!-- claim -->
|
|
430
|
<property>
|
|
431
|
<name>mapreduce.multipleoutputs.namedOutput.${out19}.key</name>
|
|
432
|
<value>org.apache.hadoop.io.Text</value>
|
|
433
|
</property>
|
|
434
|
<property>
|
|
435
|
<name>mapreduce.multipleoutputs.namedOutput.${out19}.value</name>
|
|
436
|
<value>org.apache.hadoop.io.Text</value>
|
|
437
|
</property>
|
|
438
|
<property>
|
|
439
|
<name>mapreduce.multipleoutputs.namedOutput.${out19}.format</name>
|
|
440
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
441
|
</value>
|
|
442
|
</property>
|
|
443
|
<!-- cncept -->
|
|
444
|
<property>
|
|
445
|
<name>mapreduce.multipleoutputs.namedOutput.${out20}.key</name>
|
|
446
|
<value>org.apache.hadoop.io.Text</value>
|
|
447
|
</property>
|
|
448
|
<property>
|
|
449
|
<name>mapreduce.multipleoutputs.namedOutput.${out20}.value</name>
|
|
450
|
<value>org.apache.hadoop.io.Text</value>
|
|
451
|
</property>
|
|
452
|
<property>
|
|
453
|
<name>mapreduce.multipleoutputs.namedOutput.${out20}.format</name>
|
|
454
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
455
|
</value>
|
|
456
|
</property>
|
|
457
|
|
|
458
|
<!-- category -->
|
|
459
|
<property>
|
|
460
|
<name>mapreduce.multipleoutputs.namedOutput.${out21}.key</name>
|
|
461
|
<value>org.apache.hadoop.io.Text</value>
|
|
462
|
</property>
|
|
463
|
<property>
|
|
464
|
<name>mapreduce.multipleoutputs.namedOutput.${out21}.value</name>
|
|
465
|
<value>org.apache.hadoop.io.Text</value>
|
|
466
|
</property>
|
|
467
|
<property>
|
|
468
|
<name>mapreduce.multipleoutputs.namedOutput.${out21}.format</name>
|
|
469
|
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat
|
|
470
|
</value>
|
|
471
|
</property>
|
|
472
|
<!-- ## Classes of mapper and reducer -->
|
|
473
|
|
|
474
|
<property>
|
|
475
|
<name>mapreduce.map.class</name>
|
|
476
|
<value>eu.dnetlib.data.mapreduce.hbase.statsExport.StatsMapper
|
|
477
|
</value>
|
|
478
|
</property>
|
|
479
|
<property>
|
|
480
|
<name>mapreduce.reduce.class</name>
|
|
481
|
<value>eu.dnetlib.data.mapreduce.hbase.statsExport.StatsReducer
|
|
482
|
</value>
|
|
483
|
</property>
|
|
484
|
<property>
|
|
485
|
<name>io.serializations</name>
|
|
486
|
<value>org.apache.hadoop.io.serializer.WritableSerialization
|
|
487
|
</value>
|
|
488
|
</property>
|
|
489
|
<!-- ## Custom config -->
|
|
490
|
|
|
491
|
<!--delim character used to seperate fields in hdfs dump files <property> -->
|
|
492
|
<property>
|
|
493
|
<name>stats.delim</name>
|
|
494
|
<value>${Stats_delim_Character}</value>
|
|
495
|
</property>
|
|
496
|
<!--default string for Null String Values -->
|
|
497
|
<property>
|
|
498
|
<name>stats.nullString</name>
|
|
499
|
<value>${Stats_null_String_Field}</value>
|
|
500
|
</property>
|
|
501
|
<!--default string for Null Numeric Values -->
|
|
502
|
<property>
|
|
503
|
<name>stats.nullNum</name>
|
|
504
|
<value>${Stats_null_Numeric_Field}</value>
|
|
505
|
</property>
|
|
506
|
<!--source hbase table -->
|
|
507
|
<property>
|
|
508
|
<name>hbase.mapreduce.inputtable</name>
|
|
509
|
<value>${Stats_Hbase_Source_Table}</value>
|
|
510
|
</property>
|
|
511
|
<property>
|
|
512
|
<!-- mapping of protos entities to tables in the relDB -->
|
|
513
|
<name>stats.dbTablesMap</name>
|
|
514
|
<value>${Stats_db_table_map}</value>
|
|
515
|
</property>
|
|
516
|
<property>
|
|
517
|
<name>stats.newline</name>
|
|
518
|
<value>${Stats_newline_Character}</value>
|
|
519
|
</property>
|
|
520
|
<!-- This directory does not correspond to a data store. In fact, this
|
|
521
|
directory only contains multiple data stores. It has to be set to the name
|
|
522
|
of the workflow node. -->
|
|
523
|
<property>
|
|
524
|
<name>mapred.output.dir</name>
|
|
525
|
<value>${Stats_output_Path}</value>
|
|
526
|
</property>
|
|
527
|
<property>
|
|
528
|
<name>stats.indexConf</name>
|
|
529
|
<value>${Stats_indexConf}</value>
|
|
530
|
</property>
|
|
531
|
<!-- ## Workflow node parameters -->
|
|
532
|
<property>
|
|
533
|
<name>mapred.reduce.tasks</name>
|
|
534
|
<value>${numReducers}</value>
|
|
535
|
</property>
|
|
536
|
</configuration>
|
|
537
|
</map-reduce>
|
|
538
|
<ok to="sqoopImport" />
|
|
539
|
<error to="fail" />
|
|
540
|
</action>
|
|
541
|
|
|
542
|
<action name="prepareDatabase">
|
|
543
|
<java>
|
|
544
|
|
|
545
|
<prepare>
|
|
546
|
</prepare>
|
|
547
|
<configuration>
|
|
548
|
<property>
|
|
549
|
<name>mapred.job.queue.name</name>
|
|
550
|
<value>${queueName}</value>
|
|
551
|
</property>
|
|
552
|
</configuration>
|
|
553
|
|
|
554
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
|
555
|
<arg>-SworkingDir=${workingDir}</arg>
|
|
556
|
<arg>eu.dnetlib.iis.core.workflows.stats.DBInitWrapper</arg>
|
|
557
|
|
|
558
|
<arg>-PStats_db_Url=${Stats_db_Url}</arg>
|
|
559
|
<arg>-PStats_db_User=${Stats_db_User}</arg>
|
|
560
|
<arg>-PStats_db_Pass=${Stats_db_Pass}</arg>
|
|
561
|
<arg>-PStats_db_Driver=${Stats_db_Driver}</arg>
|
|
562
|
|
|
563
|
</java>
|
|
564
|
<ok to="sqoopImport" />
|
|
565
|
<error to="fail" />
|
|
566
|
</action>
|
|
567
|
|
|
568
|
<action name="sqoopImport">
|
|
569
|
<java>
|
|
570
|
<prepare>
|
|
571
|
</prepare>
|
|
572
|
<configuration>
|
|
573
|
<property>
|
|
574
|
<name>mapred.job.queue.name</name>
|
|
575
|
<value>${queueName}</value>
|
|
576
|
</property>
|
|
577
|
|
|
578
|
<property>
|
|
579
|
<name>oozie.sqoop.log.level</name>
|
|
580
|
<value>DEBUG</value>
|
|
581
|
</property>
|
|
582
|
|
|
583
|
</configuration>
|
|
584
|
|
|
585
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
|
586
|
<arg>-SworkingDir=${workingDir}</arg>
|
|
587
|
<arg>eu.dnetlib.iis.core.workflows.stats.SqoopWrapper</arg>
|
|
588
|
|
|
589
|
<arg>-PStats_db_Url=${Stats_db_Url}</arg>
|
|
590
|
<arg>-PStats_db_User=${Stats_db_User}</arg>
|
|
591
|
<arg>-PStats_db_Pass=${Stats_db_Pass}</arg>
|
|
592
|
|
|
593
|
<arg>-PStats_output_Path=${Stats_output_Path}</arg>
|
|
594
|
<arg>-PStats_sqoop_RecsPerStatement=${Stats_sqoop_RecsPerStatement}
|
|
595
|
</arg>
|
|
596
|
<arg>-PStats_sqoop_ReducersCount=${Stats_sqoop_ReducersCount}</arg>
|
|
597
|
<arg>-PStats_sqoop_StatementPerTrans=${Stats_sqoop_StatementPerTrans}
|
|
598
|
</arg>
|
|
599
|
<arg>-PStats_delim_Character=${Stats_delim_Character}</arg>
|
|
600
|
<arg>-PStats_newline_Character=${Stats_newline_Character}</arg>
|
|
601
|
<arg>-PStats_db_table_map=${Stats_db_table_map}</arg>
|
|
602
|
|
|
603
|
|
|
604
|
|
|
605
|
|
|
606
|
</java>
|
|
607
|
<ok to="end" />
|
|
608
|
<error to="fail" />
|
|
609
|
</action>
|
|
610
|
|
|
611
|
<action name="finalizeDatabase">
|
|
612
|
<java>
|
|
613
|
|
|
614
|
<prepare>
|
|
615
|
</prepare>
|
|
616
|
<configuration>
|
|
617
|
<property>
|
|
618
|
<name>mapred.job.queue.name</name>
|
|
619
|
<value>${queueName}</value>
|
|
620
|
</property>
|
|
621
|
</configuration>
|
|
622
|
|
|
623
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
|
624
|
<arg>-SworkingDir=${workingDir}</arg>
|
|
625
|
<arg>eu.dnetlib.iis.core.workflows.stats.DBFinalizeWrapper</arg>
|
|
626
|
<arg>-PStats_db_Url=${Stats_db_Url}</arg>
|
|
627
|
<arg>-PStats_db_User=${Stats_db_User}</arg>
|
|
628
|
<arg>-PStats_db_Pass=${Stats_db_Pass}</arg>
|
|
629
|
<arg>-PStats_db_Driver=${Stats_db_Driver}</arg>
|
|
630
|
<arg>-PStats_output_Path=${Stats_output_Path}</arg>
|
|
631
|
<arg>-PStats_ContextResourceXML=${ContextResourceXML}</arg>
|
|
632
|
|
|
633
|
</java>
|
|
634
|
<ok to="end" />
|
|
635
|
<error to="fail" />
|
|
636
|
</action>
|
|
637
|
<action name="sqoopDirectImport">
|
|
638
|
<sqoop xmlns="uri:oozie:sqoop-action:0.2">
|
|
639
|
<job-tracker>${jobTracker}</job-tracker>
|
|
640
|
<name-node>${nameNode}</name-node>
|
|
641
|
<prepare>
|
|
642
|
</prepare>
|
|
643
|
<configuration>
|
|
644
|
<property>
|
|
645
|
<name>mapred.job.queue.name</name>
|
|
646
|
<value>${queueName}</value>
|
|
647
|
</property>
|
|
648
|
<property>
|
|
649
|
<name>oozie.sqoop.log.level</name>
|
|
650
|
<value>DEBUG</value>
|
|
651
|
</property>
|
|
652
|
</configuration>
|
|
653
|
<command> export -Dsqoop.export.records.per.statement=1000
|
|
654
|
-Dsqoop.statements.per.transaction==1000 --connect
|
|
655
|
jdbc:postgresql://duffy.di.uoa.gr:5432/test_stats --export-dir
|
|
656
|
/tmp/test_stats/datasource-r-00000 --table datasource --username
|
|
657
|
sqoop --password sqoop --input-fields-terminated-by ! -m 4
|
|
658
|
</command>
|
|
659
|
|
|
660
|
</sqoop>
|
|
661
|
<ok to="end" />
|
|
662
|
<error to="fail" />
|
|
663
|
</action>
|
|
664
|
|
|
665
|
<kill name="fail">
|
|
666
|
<message>
|
|
667
|
Unfortunately, the process failed -- error message:
|
|
668
|
[${wf:errorMessage(wf:lastErrorNode())}]
|
|
669
|
</message>
|
|
670
|
</kill>
|
|
671
|
<end name="end" />
|
|
672
|
</workflow-app>
|