Revision 28930
Added by Eri Katsari over 10 years ago
| modules/dnet-openaire-stats-workflow/src/main/resources/eu/dnetlib/iis/core/examples/helloTest/oozie_app/workflow.xml | ||
|---|---|---|
| 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>${queueName}</value>
|
|
| 13 |
</property> |
|
| 14 |
</configuration> |
|
| 15 |
</global> |
|
| 16 |
|
|
| 17 |
<start to='get-scanner' /> |
|
| 18 |
<action name='get-scanner'> |
|
| 19 |
<java> |
|
| 20 |
|
|
| 21 |
<main-class>eu.dnetlib.iis.core.workflows.stats.HbaseScannerGenerator |
|
| 22 |
</main-class> |
|
| 23 |
<!-- column families: --> |
|
| 24 |
|
|
| 25 |
<arg> |
|
| 26 |
-f datasource , |
|
| 27 |
datasourceOrganization_provision_provides, |
|
| 28 |
project , organization |
|
| 29 |
<!-- projectOrganization_participation_isParticipant , --> |
|
| 30 |
<!-- projectOrganization_participation_hasParticipant --> |
|
| 31 |
<!-- result, --> |
|
| 32 |
<!-- resultProject_outcome_produces , --> |
|
| 33 |
<!-- resultResult_publicationDataset_isRelatedTo, --> |
|
| 34 |
<!-- personResult_authorship_hasAuthor --> |
|
| 35 |
</arg> |
|
| 36 |
|
|
| 37 |
|
|
| 38 |
<capture-output /> |
|
| 39 |
</java> |
|
| 40 |
<ok to="mr_export" /> |
|
| 41 |
<error to="fail" /> |
|
| 42 |
</action> |
|
| 43 |
|
|
| 44 |
<action name="mr_export"> |
|
| 45 |
<map-reduce> |
|
| 46 |
|
|
| 47 |
<prepare> |
|
| 48 |
<delete path="${nameNode}${Stats_output_Path}" />
|
|
| 49 |
<!-- <delete path="${Stats_output_Path}" /> -->
|
|
| 50 |
|
|
| 51 |
</prepare> |
|
| 52 |
<configuration> |
|
| 53 |
<!-- TODO CHECK THIS --> |
|
| 54 |
<!-- HDFS --> |
|
| 55 |
<!-- hdfs://nmis-hadoop-cluster< --> |
|
| 56 |
<!-- <property> --> |
|
| 57 |
<!-- <name>dfs.nameservices</name> --> |
|
| 58 |
<!-- <value>nmis-hadoop-cluster</value> --> |
|
| 59 |
<!-- </property> --> |
|
| 60 |
|
|
| 61 |
<!-- <property> --> |
|
| 62 |
<!-- <name>dfs.ha.namenodes.nmis-hadoop-cluster</name> --> |
|
| 63 |
<!-- <value>nn1,nn2 nn1,nn2 --> |
|
| 64 |
<!-- dfs.namenode.rpc-address.nmis-hadoop-cluster.nn1=quorum1.t.hadoop.research-infrastructures.eu:8020 --> |
|
| 65 |
<!-- dfs.namenode.http-address.nmis-hadoop-cluster.nn1=quorum1.t.hadoop.research-infrastructures.eu:50070 --> |
|
| 66 |
<!-- dfs.namenode.rpc-address.nmis-hadoop-cluster.nn2=quorum2.t.hadoop.research-infrastructures.eu:8020 --> |
|
| 67 |
<!-- dfs.namenode.http-address.nmis-hadoop-cluster.nn2=quorum2.t.hadoop.research-infrastructures.eu:50070 --> |
|
| 68 |
<!-- dfs.client.failover.proxy.provider.nmis-hadoop-cluster=org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider --> |
|
| 69 |
<!-- </value> --> |
|
| 70 |
<!-- </property> --> |
|
| 71 |
|
|
| 72 |
|
|
| 73 |
<!-- HBASE --> |
|
| 74 |
<property> |
|
| 75 |
<name>hbase.mapreduce.scan</name> |
|
| 76 |
<value>${wf:actionData('get-scanner')['scan']}</value>
|
|
| 77 |
</property> |
|
| 78 |
<property> |
|
| 79 |
<name>hbase.rootdir</name> |
|
| 80 |
<value>hdfs://nmis-hadoop-cluster/hbase</value> |
|
| 81 |
</property> |
|
| 82 |
|
|
| 83 |
<property> |
|
| 84 |
<name>hbase.security.authentication</name> |
|
| 85 |
<value> simple</value> |
|
| 86 |
</property> |
|
| 87 |
<!-- ZOOKEEPER --> |
|
| 88 |
|
|
| 89 |
<property> |
|
| 90 |
<name>hbase.zookeeper.quorum</name> |
|
| 91 |
<value>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 |
|
| 92 |
</value> |
|
| 93 |
</property> |
|
| 94 |
<property> |
|
| 95 |
<name>zookeeper.znode.rootserver</name> |
|
| 96 |
<value> root-region-server |
|
| 97 |
</value> |
|
| 98 |
</property> |
|
| 99 |
|
|
| 100 |
<property> |
|
| 101 |
<name>hbase.zookeeper.property.clientPort</name> |
|
| 102 |
<value>2182</value> |
|
| 103 |
</property> |
|
| 104 |
|
|
| 105 |
<!-- MR IO --> |
|
| 106 |
|
|
| 107 |
|
|
| 108 |
<property> |
|
| 109 |
<name>mapreduce.inputformat.class</name> |
|
| 110 |
<value>org.apache.hadoop.hbase.mapreduce.TableInputFormat</value> |
|
| 111 |
</property> |
|
| 112 |
|
|
| 113 |
<property> |
|
| 114 |
<name>mapred.mapoutput.key.class</name> |
|
| 115 |
<value>org.apache.hadoop.io.Text</value> |
|
| 116 |
</property> |
|
| 117 |
<property> |
|
| 118 |
<name>mapred.mapoutput.value.class</name> |
|
| 119 |
<value>org.apache.hadoop.hbase.io.ImmutableBytesWritable</value> |
|
| 120 |
</property> |
|
| 121 |
<property> |
|
| 122 |
<name>mapred.output.key.class</name> |
|
| 123 |
<value>org.apache.hadoop.io.Text</value> |
|
| 124 |
</property> |
|
| 125 |
<property> |
|
| 126 |
<name>mapred.output.value.class</name> |
|
| 127 |
<value>org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat |
|
| 128 |
</value> |
|
| 129 |
</property> |
|
| 130 |
|
|
| 131 |
<!-- ## This is required for new MapReduce API usage --> |
|
| 132 |
<property> |
|
| 133 |
<name>mapred.mapper.new-api</name> |
|
| 134 |
<value>true</value> |
|
| 135 |
</property> |
|
| 136 |
<property> |
|
| 137 |
<name>mapred.reducer.new-api</name> |
|
| 138 |
<value>true</value> |
|
| 139 |
</property> |
|
| 140 |
|
|
| 141 |
<!-- # Job-specific options --> |
|
| 142 |
<property> |
|
| 143 |
<name>dfs.blocksize</name> |
|
| 144 |
<value>32M</value> |
|
| 145 |
</property> |
|
| 146 |
<property> |
|
| 147 |
<name>mapred.output.compress</name> |
|
| 148 |
<value>false</value> |
|
| 149 |
</property> |
|
| 150 |
<property> |
|
| 151 |
<name>mapred.reduce.tasks.speculative.execution</name> |
|
| 152 |
<value>false</value> |
|
| 153 |
</property> |
|
| 154 |
<property> |
|
| 155 |
<name>mapred.reduce.tasks.speculative.execution</name> |
|
| 156 |
<value>false</value> |
|
| 157 |
</property> |
|
| 158 |
<property> |
|
| 159 |
<name>mapreduce.map.speculative</name> |
|
| 160 |
<value>false</value> |
|
| 161 |
</property> |
|
| 162 |
|
|
| 163 |
<!-- I/O FORMAT --> |
|
| 164 |
|
|
| 165 |
<!-- ## Names of all output ports --> |
|
| 166 |
|
|
| 167 |
|
|
| 168 |
<property> |
|
| 169 |
<name>mapreduce.multipleoutputs</name> |
|
| 170 |
<value> |
|
| 171 |
datasource datasourceLanguage project organization datasourceOrganization |
|
| 172 |
<!-- datasourceTopic resultProject --> |
|
| 173 |
<!-- 'projectOrganization' --> |
|
| 174 |
<!-- result resultClaim resultClassification resultConcept --> |
|
| 175 |
<!-- resultLanguage resultOrganization resultProject resultResult resultTopic |
|
| 176 |
resultLanguage resultDatasource resultResult --> |
|
| 177 |
<!-- category claim concept --> |
|
| 178 |
</value> |
|
| 179 |
</property> |
|
| 180 |
|
|
| 181 |
<!-- datasource --> |
|
| 182 |
<property> |
|
| 183 |
<name> mapreduce.multipleoutputs.namedOutput.datasource.key</name> |
|
| 184 |
<value>org.apache.hadoop.io.Text</value> |
|
| 185 |
|
|
| 186 |
</property> |
|
| 187 |
<property> |
|
| 188 |
<name> mapreduce.multipleoutputs.namedOutput.datasource.value |
|
| 189 |
</name> |
|
| 190 |
<value>org.apache.hadoop.io.Text</value> |
|
| 191 |
</property> |
|
| 192 |
<property> |
|
| 193 |
<name> mapreduce.multipleoutputs.namedOutput.datasource.format |
|
| 194 |
</name> |
|
| 195 |
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat |
|
| 196 |
</value> |
|
| 197 |
</property> |
|
| 198 |
<!-- datasourceLanguage --> |
|
| 199 |
<property> |
|
| 200 |
<name>mapreduce.multipleoutputs.namedOutput.datasourceLanguage.key |
|
| 201 |
</name> |
|
| 202 |
<value>org.apache.hadoop.io.Text</value> |
|
| 203 |
</property> |
|
| 204 |
<property> |
|
| 205 |
|
|
| 206 |
<name> |
|
| 207 |
mapreduce.multipleoutputs.namedOutput.datasourceLanguage.value |
|
| 208 |
</name> |
|
| 209 |
<value>org.apache.hadoop.io.Text</value> |
|
| 210 |
</property> |
|
| 211 |
<property> |
|
| 212 |
<name> |
|
| 213 |
mapreduce.multipleoutputs.namedOutput.datasourceLanguage.format |
|
| 214 |
</name> |
|
| 215 |
<value> |
|
| 216 |
org.apache.hadoop.mapreduce.lib.output.TextOutputFormat |
|
| 217 |
|
|
| 218 |
</value> |
|
| 219 |
</property> |
|
| 220 |
|
|
| 221 |
<!-- datasourceOrganization --> |
|
| 222 |
<property> |
|
| 223 |
<name>mapreduce.multipleoutputs.namedOutput.datasourceOrganization.key |
|
| 224 |
</name> |
|
| 225 |
<value>org.apache.hadoop.io.Text</value> |
|
| 226 |
</property> |
|
| 227 |
<property> |
|
| 228 |
|
|
| 229 |
<name> |
|
| 230 |
mapreduce.multipleoutputs.namedOutput.datasourceOrganization.value |
|
| 231 |
</name> |
|
| 232 |
<value>org.apache.hadoop.io.Text</value> |
|
| 233 |
</property> |
|
| 234 |
<property> |
|
| 235 |
<name> |
|
| 236 |
mapreduce.multipleoutputs.namedOutput.datasourceOrganization.format |
|
| 237 |
</name> |
|
| 238 |
<value> |
|
| 239 |
org.apache.hadoop.mapreduce.lib.output.TextOutputFormat |
|
| 240 |
|
|
| 241 |
</value> |
|
| 242 |
</property> |
|
| 243 |
|
|
| 244 |
<!-- datasourceTopic --> |
|
| 245 |
<!-- <property> --> |
|
| 246 |
<!-- <name>mapreduce.multipleoutputs.namedOutput.datasourceTopic.key --> |
|
| 247 |
<!-- </name> --> |
|
| 248 |
<!-- <value>org.apache.hadoop.io.Text</value> --> |
|
| 249 |
<!-- </property> --> |
|
| 250 |
<!-- <property> --> |
|
| 251 |
|
|
| 252 |
<!-- <name> --> |
|
| 253 |
<!-- mapreduce.multipleoutputs.namedOutput.datasourceTopic.value --> |
|
| 254 |
<!-- </name> --> |
|
| 255 |
<!-- <value>org.apache.hadoop.io.Text</value> --> |
|
| 256 |
<!-- </property> --> |
|
| 257 |
<!-- <property> --> |
|
| 258 |
<!-- <name> --> |
|
| 259 |
<!-- mapreduce.multipleoutputs.namedOutput.datasourceTopic.format --> |
|
| 260 |
<!-- </name> --> |
|
| 261 |
<!-- <value> --> |
|
| 262 |
<!-- org.apache.hadoop.mapreduce.lib.output.TextOutputFormat --> |
|
| 263 |
|
|
| 264 |
<!-- </value> --> |
|
| 265 |
<!-- </property> --> |
|
| 266 |
|
|
| 267 |
<!-- project --> |
|
| 268 |
<property> |
|
| 269 |
<name>mapreduce.multipleoutputs.namedOutput.project.key |
|
| 270 |
</name> |
|
| 271 |
<value>org.apache.hadoop.io.Text</value> |
|
| 272 |
</property> |
|
| 273 |
<property> |
|
| 274 |
|
|
| 275 |
<name> |
|
| 276 |
mapreduce.multipleoutputs.namedOutput.project.value |
|
| 277 |
</name> |
|
| 278 |
<value>org.apache.hadoop.io.Text</value> |
|
| 279 |
</property> |
|
| 280 |
<property> |
|
| 281 |
<name> |
|
| 282 |
mapreduce.multipleoutputs.namedOutput.project.format |
|
| 283 |
</name> |
|
| 284 |
<value> |
|
| 285 |
org.apache.hadoop.mapreduce.lib.output.TextOutputFormat |
|
| 286 |
|
|
| 287 |
</value> |
|
| 288 |
</property> |
|
| 289 |
|
|
| 290 |
<!-- resultProject --> |
|
| 291 |
<!-- <property> --> |
|
| 292 |
<!-- <name>mapreduce.multipleoutputs.namedOutput.resultProject.key --> |
|
| 293 |
<!-- </name> --> |
|
| 294 |
<!-- <value>org.apache.hadoop.io.Text</value> --> |
|
| 295 |
<!-- </property> --> |
|
| 296 |
<!-- <property> --> |
|
| 297 |
|
|
| 298 |
<!-- <name> --> |
|
| 299 |
<!-- mapreduce.multipleoutputs.namedOutput.resultProject.value --> |
|
| 300 |
<!-- </name> --> |
|
| 301 |
<!-- <value>org.apache.hadoop.io.Text</value> --> |
|
| 302 |
<!-- </property> --> |
|
| 303 |
<!-- <property> --> |
|
| 304 |
<!-- <name> --> |
|
| 305 |
<!-- mapreduce.multipleoutputs.namedOutput.resultProject.format --> |
|
| 306 |
<!-- </name> --> |
|
| 307 |
<!-- <value> --> |
|
| 308 |
<!-- org.apache.hadoop.mapreduce.lib.output.TextOutputFormat --> |
|
| 309 |
|
|
| 310 |
<!-- </value> --> |
|
| 311 |
<!-- </property> --> |
|
| 312 |
<!-- projectOrganization --> |
|
| 313 |
<!-- <property> --> |
|
| 314 |
<!-- <name>mapreduce.multipleoutputs.namedOutput.projectOrganization.key --> |
|
| 315 |
<!-- </name> --> |
|
| 316 |
<!-- <value>org.apache.hadoop.io.Text</value> --> |
|
| 317 |
<!-- </property> --> |
|
| 318 |
<!-- <property> --> |
|
| 319 |
|
|
| 320 |
<!-- <name> --> |
|
| 321 |
<!-- mapreduce.multipleoutputs.namedOutput.projectOrganization.value --> |
|
| 322 |
<!-- </name> --> |
|
| 323 |
<!-- <value>org.apache.hadoop.io.Text</value> --> |
|
| 324 |
<!-- </property> --> |
|
| 325 |
<!-- <property> --> |
|
| 326 |
<!-- <name> --> |
|
| 327 |
<!-- mapreduce.multipleoutputs.namedOutput.projectOrganization.format --> |
|
| 328 |
<!-- </name> --> |
|
| 329 |
<!-- <value> --> |
|
| 330 |
<!-- org.apache.hadoop.mapreduce.lib.output.TextOutputFormat --> |
|
| 331 |
|
|
| 332 |
<!-- </value> --> |
|
| 333 |
<!-- </property> --> |
|
| 334 |
|
|
| 335 |
|
|
| 336 |
<!-- organization --> |
|
| 337 |
<property> |
|
| 338 |
<name>mapreduce.multipleoutputs.namedOutput.organization.key |
|
| 339 |
</name> |
|
| 340 |
<value>org.apache.hadoop.io.Text</value> |
|
| 341 |
</property> |
|
| 342 |
<property> |
|
| 343 |
|
|
| 344 |
<name> |
|
| 345 |
mapreduce.multipleoutputs.namedOutput.organization.value |
|
| 346 |
</name> |
|
| 347 |
<value>org.apache.hadoop.io.Text</value> |
|
| 348 |
</property> |
|
| 349 |
<property> |
|
| 350 |
<name> |
|
| 351 |
mapreduce.multipleoutputs.namedOutput.organization.format |
|
| 352 |
</name> |
|
| 353 |
<value> |
|
| 354 |
org.apache.hadoop.mapreduce.lib.output.TextOutputFormat |
|
| 355 |
|
|
| 356 |
</value> |
|
| 357 |
</property> |
|
| 358 |
|
|
| 359 |
|
|
| 360 |
<!-- ## Classes of mapper and reducer --> |
|
| 361 |
|
|
| 362 |
<property> |
|
| 363 |
<name>mapreduce.map.class</name> |
|
| 364 |
<value>eu.dnetlib.data.mapreduce.hbase.statsExport.StatsMapper |
|
| 365 |
</value> |
|
| 366 |
</property> |
|
| 367 |
<property> |
|
| 368 |
<name>mapreduce.reduce.class</name> |
|
| 369 |
<value>eu.dnetlib.data.mapreduce.hbase.statsExport.StatsReducer |
|
| 370 |
</value> |
|
| 371 |
</property> |
|
| 372 |
<property> |
|
| 373 |
<name>io.serializations</name> |
|
| 374 |
<value>org.apache.hadoop.io.serializer.WritableSerialization |
|
| 375 |
</value> |
|
| 376 |
</property> |
|
| 377 |
<!-- ## Custom config --> |
|
| 378 |
|
|
| 379 |
<!--delim character used to seperate fields in hdfs dump files <property> --> |
|
| 380 |
<property> |
|
| 381 |
<name>stats.delim </name> |
|
| 382 |
<value>${Stats_delim_Character}</value>
|
|
| 383 |
</property> |
|
| 384 |
<!--default string for Null String Values --> |
|
| 385 |
<property> |
|
| 386 |
<name>stats.nullString</name> |
|
| 387 |
<value>${Stats_null_String_Field}</value>
|
|
| 388 |
</property> |
|
| 389 |
<!--default string for Null Numeric Values --> |
|
| 390 |
<property> |
|
| 391 |
<name>stats.nullNum</name> |
|
| 392 |
<value>${Stats_null_Numeric_Field}</value>
|
|
| 393 |
</property> |
|
| 394 |
<!--source hbase table --> |
|
| 395 |
<property> |
|
| 396 |
<name>hbase.mapreduce.inputtable</name> |
|
| 397 |
<value>${Stats_Hbase_Source_Table}</value>
|
|
| 398 |
</property> |
|
| 399 |
<!-- This directory does not correspond to a data store. In fact, this |
|
| 400 |
directory only contains multiple data stores. It has to be set to the name |
|
| 401 |
of the workflow node. --> |
|
| 402 |
<property> |
|
| 403 |
<name>mapred.output.dir</name> |
|
| 404 |
<value>${Stats_output_Path}</value>
|
|
| 405 |
</property> |
|
| 406 |
<!-- ## Workflow node parameters --> |
|
| 407 |
<property> |
|
| 408 |
<name>mapred.reduce.tasks</name> |
|
| 409 |
<value>1</value> |
|
| 410 |
</property> |
|
| 411 |
</configuration> |
|
| 412 |
</map-reduce> |
|
| 413 |
<ok to="end" /> |
|
| 414 |
<error to="fail" /> |
|
| 415 |
</action> |
|
| 416 |
|
|
| 417 |
|
|
| 418 |
<action name="prepareDatabase"> |
|
| 419 |
<java> |
|
| 420 |
|
|
| 421 |
<prepare> |
|
| 422 |
</prepare> |
|
| 423 |
<configuration> |
|
| 424 |
<property> |
|
| 425 |
<name>mapred.job.queue.name</name> |
|
| 426 |
<value>${queueName}</value>
|
|
| 427 |
</property> |
|
| 428 |
</configuration> |
|
| 429 |
|
|
| 430 |
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper |
|
| 431 |
</main-class> |
|
| 432 |
<arg>-SworkingDir=${workingDir}</arg>
|
|
| 433 |
<arg>eu.dnetlib.iis.core.workflows.stats.DBInitWrapper</arg> |
|
| 434 |
|
|
| 435 |
<arg>-PStats_db_Url=${Stats_db_Url}</arg>
|
|
| 436 |
<arg>-PStats_db_User=${Stats_db_User}</arg>
|
|
| 437 |
<arg>-PStats_db_Pass=${Stats_db_Pass}</arg>
|
|
| 438 |
<arg>-PStats_db_Name=${Stats_db_Name}</arg>
|
|
| 439 |
<arg>-PStats_db_Driver=${Stats_db_Driver}</arg>
|
|
| 440 |
|
|
| 441 |
</java> |
|
| 442 |
<ok to="finalizeDatabase" /> |
|
| 443 |
<error to="fail" /> |
|
| 444 |
</action> |
|
| 445 |
|
|
| 446 |
<action name="sqoopImport"> |
|
| 447 |
<java> |
|
| 448 |
<prepare> |
|
| 449 |
</prepare> |
|
| 450 |
<configuration> |
|
| 451 |
<property> |
|
| 452 |
<name>mapred.job.queue.name</name> |
|
| 453 |
<value>${queueName}</value>
|
|
| 454 |
</property> |
|
| 455 |
</configuration> |
|
| 456 |
|
|
| 457 |
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper |
|
| 458 |
</main-class> |
|
| 459 |
<arg>-SworkingDir=${workingDir}</arg>
|
|
| 460 |
<arg>eu.dnetlib.iis.core.workflows.stats.SqoopWrapper</arg> |
|
| 461 |
|
|
| 462 |
<arg>-PStats_db_Url=${Stats_db_Url}</arg>
|
|
| 463 |
<arg>-PStats_db_User=${Stats_db_User}</arg>
|
|
| 464 |
<arg>-PStats_db_Pass=${Stats_db_Pass}</arg>
|
|
| 465 |
<arg>-PStats_db_Name=${Stats_db_Name}</arg>
|
|
| 466 |
<arg>-PStats_output_Path=${Stats_output_Path}</arg>
|
|
| 467 |
<arg>-PStats_sqoop_RecsPerStatement=${Stats_sqoop_RecsPerStatement}
|
|
| 468 |
</arg> |
|
| 469 |
<arg>-PStats_sqoop_ReducersCount=${Stats_sqoop_ReducersCount}</arg>
|
|
| 470 |
<arg>-PStats_sqoop_StatementPerTrans=${Stats_sqoop_StatementPerTrans}
|
|
| 471 |
</arg> |
|
| 472 |
<!-- <java-opts>-Dlog4j.configuration=log4jConfig</java-opts> --> |
|
| 473 |
</java> |
|
| 474 |
<ok to="end" /> |
|
| 475 |
<error to="fail" /> |
|
| 476 |
</action> |
|
| 477 |
|
|
| 478 |
<action name="sqoopDirectImport"> |
|
| 479 |
<sqoop xmlns="uri:oozie:sqoop-action:0.2"> |
|
| 480 |
<job-tracker>${jobTracker}</job-tracker>
|
|
| 481 |
<name-node>${nameNode}</name-node>
|
|
| 482 |
<prepare> |
|
| 483 |
|
|
| 484 |
|
|
| 485 |
</prepare> |
|
| 486 |
<configuration> |
|
| 487 |
<property> |
|
| 488 |
<name>mapred.compress.map.output</name> |
|
| 489 |
<value>true</value> |
|
| 490 |
</property> |
|
| 491 |
</configuration> |
|
| 492 |
<arg>export</arg> |
|
| 493 |
<arg> -Dsqoop.export.records.per.statement</arg> |
|
| 494 |
<arg> ${Stats_sqoop_RecsPerStatement}</arg>
|
|
| 495 |
|
|
| 496 |
|
|
| 497 |
<arg>-Dsqoop.export.statements.per.transaction </arg> |
|
| 498 |
<arg>${Stats_sqoop_StatementPerTrans}</arg>
|
|
| 499 |
|
|
| 500 |
|
|
| 501 |
<arg>--connect</arg> |
|
| 502 |
<arg>${Stats_db_Url}/${Stats_db_Name}</arg>
|
|
| 503 |
<arg>--table</arg> |
|
| 504 |
<arg>datasource</arg> |
|
| 505 |
|
|
| 506 |
<arg>--table</arg> |
|
| 507 |
<arg>datasource</arg> |
|
| 508 |
|
|
| 509 |
<arg>--export-dir</arg> |
|
| 510 |
<arg>${Stats_output_Path}/datasource-r-00000</arg>
|
|
| 511 |
<arg> --optionally-enclosed-by </arg> |
|
| 512 |
<arg> " " </arg> |
|
| 513 |
<arg>--input-fields-terminated-by </arg> |
|
| 514 |
<arg> ${Stats_delim_Character } </arg>
|
|
| 515 |
<arg> --verbose </arg> |
|
| 516 |
<arg>--username </arg> |
|
| 517 |
<arg>${Stats_db_User} </arg>
|
|
| 518 |
<arg>--password</arg> |
|
| 519 |
<arg>${Stats_db_Pass} </arg>
|
|
| 520 |
<arg>--batch</arg> |
|
| 521 |
<arg>-m</arg> |
|
| 522 |
<arg>${Stats_sqoop_ReducersCount}</arg>
|
|
| 523 |
</sqoop> |
|
| 524 |
<ok to="end" /> |
|
| 525 |
<error to="fail" /> |
|
| 526 |
</action> |
|
| 527 |
<action name="finalizeDatabase"> |
|
| 528 |
<java> |
|
| 529 |
|
|
| 530 |
<prepare> |
|
| 531 |
</prepare> |
|
| 532 |
<configuration> |
|
| 533 |
<property> |
|
| 534 |
<name>mapred.job.queue.name</name> |
|
| 535 |
<value>${queueName}</value>
|
|
| 536 |
</property> |
|
| 537 |
</configuration> |
|
| 538 |
|
|
| 539 |
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper |
|
| 540 |
</main-class> |
|
| 541 |
<arg>-SworkingDir=${workingDir}</arg>
|
|
| 542 |
<arg>eu.dnetlib.iis.core.workflows.stats.DBFinalizeWrapper |
|
| 543 |
</arg> |
|
| 544 |
<arg>-PStats_db_Url=${Stats_db_Url}</arg>
|
|
| 545 |
<arg>-PStats_db_User=${Stats_db_User}</arg>
|
|
| 546 |
<arg>-PStats_db_Pass=${Stats_db_Pass}</arg>
|
|
| 547 |
<arg>-PStats_db_Name=${Stats_db_Name}</arg>
|
|
| 548 |
<arg>-PStats_db_Driver=${Stats_db_Driver}</arg>
|
|
| 549 |
</java> |
|
| 550 |
<ok to="end" /> |
|
| 551 |
<error to="fail" /> |
|
| 552 |
</action> |
|
| 553 |
|
|
| 554 |
|
|
| 555 |
<kill name="fail"> |
|
| 556 |
<message> |
|
| 557 |
Unfortunately, the process failed -- error message: |
|
| 558 |
[${wf:errorMessage(wf:lastErrorNode())}]
|
|
| 559 |
</message> |
|
| 560 |
</kill> |
|
| 561 |
<end name="end" /> |
|
| 562 |
</workflow-app> |
|
| 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 to the relation |
|
| 4 |
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>${queueName}</value>
|
|
| 13 |
</property> |
|
| 14 |
</configuration> |
|
| 15 |
</global> |
|
| 16 |
<start to='sqoopDirectImport' /> |
|
| 17 |
<action name='get-scanner'> |
|
| 18 |
<java> |
|
| 19 |
|
|
| 20 |
<main-class>eu.dnetlib.iis.core.workflows.stats.HbaseScannerGenerator</main-class> |
|
| 21 |
<!-- column families: --> |
|
| 22 |
|
|
| 23 |
<arg> |
|
| 24 |
-f datasource |
|
| 25 |
<!--, datasourceOrganization_provision_provides, project , organization--> |
|
| 26 |
<!-- projectOrganization_participation_isParticipant , --> |
|
| 27 |
<!-- projectOrganization_participation_hasParticipant --> |
|
| 28 |
<!-- result, --> |
|
| 29 |
<!-- resultProject_outcome_produces , --> |
|
| 30 |
<!-- resultResult_publicationDataset_isRelatedTo, --> |
|
| 31 |
<!-- personResult_authorship_hasAuthor --> |
|
| 32 |
</arg> |
|
| 33 |
|
|
| 34 |
|
|
| 35 |
<capture-output /> |
|
| 36 |
</java> |
|
| 37 |
<ok to="mr_export" /> |
|
| 38 |
<error to="fail" /> |
|
| 39 |
</action> |
|
| 40 |
<action name='get-multiple_outputs'> |
|
| 41 |
<java> |
|
| 42 |
|
|
| 43 |
<main-class>eu.dnetlib.iis.core.workflows.stats.MultipleOutputsGenerator</main-class> |
|
| 44 |
<!-- MoS names --> |
|
| 45 |
|
|
| 46 |
<arg>-n datasource , |
|
| 47 |
datasourceLanguage</arg> |
|
| 48 |
|
|
| 49 |
<arg>-k org.apache.hadoop.io.Text.class</arg> |
|
| 50 |
|
|
| 51 |
<arg>-v org.apache.hadoop.io.Text.class</arg> |
|
| 52 |
|
|
| 53 |
|
|
| 54 |
<arg>-f org.apache.hadoop.mapreduce.lib.output.TextOutputFormat.class</arg> |
|
| 55 |
|
|
| 56 |
<capture-output /> |
|
| 57 |
</java> |
|
| 58 |
<ok to="mr_export" /> |
|
| 59 |
<error to="fail" /> |
|
| 60 |
</action> |
|
| 61 |
<action name="mr_export"> |
|
| 62 |
<map-reduce> |
|
| 63 |
|
|
| 64 |
<prepare> |
|
| 65 |
<delete path="${nameNode}${Stats_output_Path}" />
|
|
| 66 |
|
|
| 67 |
|
|
| 68 |
</prepare> |
|
| 69 |
<configuration> |
|
| 70 |
<!-- TODO CHECK THIS --> |
|
| 71 |
<!-- HDFS --> |
|
| 72 |
<!-- hdfs://nmis-hadoop-cluster< --> |
|
| 73 |
<!-- <property> --> |
|
| 74 |
<!-- <name>dfs.nameservices</name> --> |
|
| 75 |
<!-- <value>nmis-hadoop-cluster</value> --> |
|
| 76 |
<!-- </property> --> |
|
| 77 |
|
|
| 78 |
<!-- <property> --> |
|
| 79 |
<!-- <name>dfs.ha.namenodes.nmis-hadoop-cluster</name> --> |
|
| 80 |
<!-- <value>nn1,nn2 nn1,nn2 --> |
|
| 81 |
<!-- dfs.namenode.rpc-address.nmis-hadoop-cluster.nn1=quorum1.t.hadoop.research-infrastructures.eu:8020 |
|
| 82 |
--> |
|
| 83 |
<!-- dfs.namenode.http-address.nmis-hadoop-cluster.nn1=quorum1.t.hadoop.research-infrastructures.eu:50070 |
|
| 84 |
--> |
|
| 85 |
<!-- dfs.namenode.rpc-address.nmis-hadoop-cluster.nn2=quorum2.t.hadoop.research-infrastructures.eu:8020 |
|
| 86 |
--> |
|
| 87 |
<!-- dfs.namenode.http-address.nmis-hadoop-cluster.nn2=quorum2.t.hadoop.research-infrastructures.eu:50070 |
|
| 88 |
--> |
|
| 89 |
<!-- dfs.client.failover.proxy.provider.nmis-hadoop-cluster=org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider |
|
| 90 |
--> |
|
| 91 |
<!-- </value> --> |
|
| 92 |
<!-- </property> --> |
|
| 93 |
|
|
| 94 |
|
|
| 95 |
<!-- HBASE --> |
|
| 96 |
<property> |
|
| 97 |
<name>hbase.mapreduce.scan</name> |
|
| 98 |
<value>${wf:actionData('get-scanner')['scan']}</value>
|
|
| 99 |
</property> |
|
| 100 |
<property> |
|
| 101 |
<name>hbase.rootdir</name> |
|
| 102 |
<value>hdfs://nmis-hadoop-cluster/hbase</value> |
|
| 103 |
</property> |
|
| 104 |
|
|
| 105 |
<property> |
|
| 106 |
<name>hbase.security.authentication</name> |
|
| 107 |
<value>simple</value> |
|
| 108 |
</property> |
|
| 109 |
<!-- ZOOKEEPER --> |
|
| 110 |
|
|
| 111 |
<property> |
|
| 112 |
<name>hbase.zookeeper.quorum</name> |
|
| 113 |
<value> |
|
| 114 |
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 |
|
| 115 |
</value> |
|
| 116 |
</property> |
|
| 117 |
<property> |
|
| 118 |
<name>zookeeper.znode.rootserver</name> |
|
| 119 |
<value>root-region-server</value> |
|
| 120 |
</property> |
|
| 121 |
|
|
| 122 |
<property> |
|
| 123 |
<name>hbase.zookeeper.property.clientPort</name> |
|
| 124 |
<value>2182</value> |
|
| 125 |
</property> |
|
| 126 |
|
|
| 127 |
|
|
| 128 |
<property> |
|
| 129 |
<name>mapreduce.framework.name</name> |
|
| 130 |
<value>classic</value> |
|
| 131 |
</property> |
|
| 132 |
|
|
| 133 |
<property> |
|
| 134 |
<name>mapreduce.jobtracker.address</name> |
|
| 135 |
<value>nmis-hadoop-jt</value> |
|
| 136 |
</property> |
|
| 137 |
|
|
| 138 |
<!-- MR IO --> |
|
| 139 |
|
|
| 140 |
|
|
| 141 |
<property> |
|
| 142 |
<name>mapreduce.inputformat.class</name> |
|
| 143 |
<value>org.apache.hadoop.hbase.mapreduce.TableInputFormat</value> |
|
| 144 |
</property> |
|
| 145 |
|
|
| 146 |
<property> |
|
| 147 |
<name>mapred.mapoutput.key.class</name> |
|
| 148 |
<value>org.apache.hadoop.io.Text</value> |
|
| 149 |
</property> |
|
| 150 |
<property> |
|
| 151 |
<name>mapred.mapoutput.value.class</name> |
|
| 152 |
<value>org.apache.hadoop.hbase.io.ImmutableBytesWritable</value> |
|
| 153 |
</property> |
|
| 154 |
<property> |
|
| 155 |
<name>mapred.output.key.class</name> |
|
| 156 |
<value>org.apache.hadoop.io.Text</value> |
|
| 157 |
</property> |
|
| 158 |
<property> |
|
| 159 |
<name>mapred.output.value.class</name> |
|
| 160 |
<value>org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat</value> |
|
| 161 |
</property> |
|
| 162 |
|
|
| 163 |
<!-- ## This is required for new MapReduce API usage --> |
|
| 164 |
<property> |
|
| 165 |
<name>mapred.mapper.new-api</name> |
|
| 166 |
<value>true</value> |
|
| 167 |
</property> |
|
| 168 |
<property> |
|
| 169 |
<name>mapred.reducer.new-api</name> |
|
| 170 |
<value>true</value> |
|
| 171 |
</property> |
|
| 172 |
|
|
| 173 |
<!-- # Job-specific options --> |
|
| 174 |
<property> |
|
| 175 |
<name>dfs.blocksize</name> |
|
| 176 |
<value>32M</value> |
|
| 177 |
</property> |
|
| 178 |
<property> |
|
| 179 |
<name>mapred.output.compress</name> |
|
| 180 |
<value>false</value> |
|
| 181 |
</property> |
|
| 182 |
<property> |
|
| 183 |
<name>mapred.reduce.tasks.speculative.execution</name> |
|
| 184 |
<value>false</value> |
|
| 185 |
</property> |
|
| 186 |
<property> |
|
| 187 |
<name>mapred.reduce.tasks.speculative.execution</name> |
|
| 188 |
<value>false</value> |
|
| 189 |
</property> |
|
| 190 |
<property> |
|
| 191 |
<name>mapreduce.map.speculative</name> |
|
| 192 |
<value>false</value> |
|
| 193 |
</property> |
|
| 194 |
|
|
| 195 |
<!-- I/O FORMAT --> |
|
| 196 |
|
|
| 197 |
<!-- ## Names of all output ports --> |
|
| 198 |
|
|
| 199 |
<property> |
|
| 200 |
<name>mapreduce.multipleoutputs</name> |
|
| 201 |
|
|
| 202 |
<value> |
|
| 203 |
datasource datasourceLanguage |
|
| 204 |
<!-- project organization datasourceOrganization--> |
|
| 205 |
<!-- datasourceTopic resultProject --> |
|
| 206 |
<!-- 'projectOrganization' --> |
|
| 207 |
<!-- result resultClaim resultClassification resultConcept --> |
|
| 208 |
<!-- resultLanguage resultOrganization resultProject resultResult resultTopic resultLanguage |
|
| 209 |
resultDatasource resultResult --> |
|
| 210 |
<!-- category claim concept --> |
|
| 211 |
</value> |
|
| 212 |
</property> |
|
| 213 |
<!-- datasource --> |
|
| 214 |
<property> |
|
| 215 |
<name>mapreduce.multipleoutputs.namedOutput.datasource.key</name> |
|
| 216 |
<value>org.apache.hadoop.io.Text</value> |
|
| 217 |
</property> |
|
| 218 |
<property> |
|
| 219 |
<name>mapreduce.multipleoutputs.namedOutput.datasource.value</name> |
|
| 220 |
<value>org.apache.hadoop.io.Text</value> |
|
| 221 |
</property> |
|
| 222 |
<property> |
|
| 223 |
<name>mapreduce.multipleoutputs.namedOutput.datasource.format</name> |
|
| 224 |
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat</value> |
|
| 225 |
</property> |
|
| 226 |
<!-- datasourceLanguage --> |
|
| 227 |
<property> |
|
| 228 |
<name>mapreduce.multipleoutputs.namedOutput.datasourceLanguage.key</name> |
|
| 229 |
<value>org.apache.hadoop.io.Text</value> |
|
| 230 |
</property> |
|
| 231 |
<property> |
|
| 232 |
<name>mapreduce.multipleoutputs.namedOutput.datasourceLanguage.value</name> |
|
| 233 |
<value>org.apache.hadoop.io.Text</value> |
|
| 234 |
</property> |
|
| 235 |
<property> |
|
| 236 |
<name>mapreduce.multipleoutputs.namedOutput.datasourceLanguage.format</name> |
|
| 237 |
<value>org.apache.hadoop.mapreduce.lib.output.TextOutputFormat</value> |
|
| 238 |
</property> |
|
| 239 |
|
|
| 240 |
<!-- datasourceOrganization --> |
|
| 241 |
|
|
| 242 |
<!--<property> <name>mapreduce.multipleoutputs.namedOutput.datasourceOrganization.key |
|
| 243 |
</name> <value>org.apache.hadoop.io.Text</value> </property> <property> <name> |
|
| 244 |
mapreduce.multipleoutputs.namedOutput.datasourceOrganization.value </name> <value>org.apache.hadoop.io.Text</value> |
|
| 245 |
</property> <property> <name> mapreduce.multipleoutputs.namedOutput.datasourceOrganization.format |
|
| 246 |
</name> <value> org.apache.hadoop.mapreduce.lib.output.TextOutputFormat </value> |
|
| 247 |
</property> --> |
|
| 248 |
<!-- datasourceTopic --> |
|
| 249 |
<!-- <property> --> |
|
| 250 |
<!-- <name>mapreduce.multipleoutputs.namedOutput.datasourceTopic.key --> |
|
| 251 |
<!-- </name> --> |
|
| 252 |
<!-- <value>org.apache.hadoop.io.Text</value> --> |
|
| 253 |
<!-- </property> --> |
|
| 254 |
<!-- <property> --> |
|
| 255 |
|
|
| 256 |
<!-- <name> --> |
|
| 257 |
<!-- mapreduce.multipleoutputs.namedOutput.datasourceTopic.value --> |
|
| 258 |
<!-- </name> --> |
|
| 259 |
<!-- <value>org.apache.hadoop.io.Text</value> --> |
|
| 260 |
<!-- </property> --> |
|
| 261 |
<!-- <property> --> |
|
| 262 |
<!-- <name> --> |
|
| 263 |
<!-- mapreduce.multipleoutputs.namedOutput.datasourceTopic.format --> |
|
| 264 |
<!-- </name> --> |
|
| 265 |
<!-- <value> --> |
|
| 266 |
<!-- org.apache.hadoop.mapreduce.lib.output.TextOutputFormat --> |
|
| 267 |
|
|
| 268 |
<!-- </value> --> |
|
| 269 |
<!-- </property> --> |
|
| 270 |
|
|
| 271 |
<!-- project --> |
|
| 272 |
<!--<property> <name>mapreduce.multipleoutputs.namedOutput.project.key </name> <value>org.apache.hadoop.io.Text</value> |
|
| 273 |
</property> <property> <name> mapreduce.multipleoutputs.namedOutput.project.value |
|
| 274 |
</name> <value>org.apache.hadoop.io.Text</value> </property> <property> <name> |
|
| 275 |
mapreduce.multipleoutputs.namedOutput.project.format </name> <value> org.apache.hadoop.mapreduce.lib.output.TextOutputFormat |
|
| 276 |
</value> </property> --> |
|
| 277 |
<!-- resultProject --> |
|
| 278 |
<!-- <property> --> |
|
| 279 |
<!-- <name>mapreduce.multipleoutputs.namedOutput.resultProject.key --> |
|
| 280 |
<!-- </name> --> |
|
| 281 |
<!-- <value>org.apache.hadoop.io.Text</value> --> |
|
| 282 |
<!-- </property> --> |
|
| 283 |
<!-- <property> --> |
|
| 284 |
|
|
| 285 |
<!-- <name> --> |
|
| 286 |
<!-- mapreduce.multipleoutputs.namedOutput.resultProject.value --> |
|
| 287 |
<!-- </name> --> |
|
| 288 |
<!-- <value>org.apache.hadoop.io.Text</value> --> |
|
| 289 |
<!-- </property> --> |
|
| 290 |
<!-- <property> --> |
|
| 291 |
<!-- <name> --> |
|
| 292 |
<!-- mapreduce.multipleoutputs.namedOutput.resultProject.format --> |
|
| 293 |
<!-- </name> --> |
|
| 294 |
<!-- <value> --> |
|
| 295 |
<!-- org.apache.hadoop.mapreduce.lib.output.TextOutputFormat --> |
|
| 296 |
|
|
| 297 |
<!-- </value> --> |
|
| 298 |
<!-- </property> --> |
|
| 299 |
<!-- projectOrganization --> |
|
| 300 |
<!-- <property> --> |
|
| 301 |
<!-- <name>mapreduce.multipleoutputs.namedOutput.projectOrganization.key --> |
|
| 302 |
<!-- </name> --> |
|
| 303 |
<!-- <value>org.apache.hadoop.io.Text</value> --> |
|
| 304 |
<!-- </property> --> |
|
| 305 |
<!-- <property> --> |
|
| 306 |
|
|
| 307 |
<!-- <name> --> |
|
| 308 |
<!-- mapreduce.multipleoutputs.namedOutput.projectOrganization.value --> |
|
| 309 |
<!-- </name> --> |
|
| 310 |
<!-- <value>org.apache.hadoop.io.Text</value> --> |
|
| 311 |
<!-- </property> --> |
|
| 312 |
<!-- <property> --> |
|
| 313 |
<!-- <name> --> |
|
| 314 |
<!-- mapreduce.multipleoutputs.namedOutput.projectOrganization.format --> |
|
| 315 |
<!-- </name> --> |
|
| 316 |
<!-- <value> --> |
|
| 317 |
<!-- org.apache.hadoop.mapreduce.lib.output.TextOutputFormat --> |
|
| 318 |
|
|
| 319 |
<!-- </value> --> |
|
| 320 |
<!-- </property> --> |
|
| 321 |
|
|
| 322 |
|
|
| 323 |
<!-- organization <property> <name>mapreduce.multipleoutputs.namedOutput.organization.key |
|
| 324 |
</name> <value>org.apache.hadoop.io.Text</value> </property> <property> <name> |
|
| 325 |
mapreduce.multipleoutputs.namedOutput.organization.value </name> <value>org.apache.hadoop.io.Text</value> |
|
| 326 |
</property> <property> <name> mapreduce.multipleoutputs.namedOutput.organization.format |
|
| 327 |
</name> <value> org.apache.hadoop.mapreduce.lib.output.TextOutputFormat </value> |
|
| 328 |
</property> --> |
|
| 329 |
|
|
| 330 |
<!-- ## Classes of mapper and reducer --> |
|
| 331 |
|
|
| 332 |
<property> |
|
| 333 |
<name>mapreduce.map.class</name> |
|
| 334 |
<value>eu.dnetlib.data.mapreduce.hbase.statsExport.StatsMapper</value> |
|
| 335 |
</property> |
|
| 336 |
<property> |
|
| 337 |
<name>mapreduce.reduce.class</name> |
|
| 338 |
<value>eu.dnetlib.data.mapreduce.hbase.statsExport.StatsReducer</value> |
|
| 339 |
</property> |
|
| 340 |
<property> |
|
| 341 |
<name>io.serializations</name> |
|
| 342 |
<value>org.apache.hadoop.io.serializer.WritableSerialization</value> |
|
| 343 |
</property> |
|
| 344 |
<!-- ## Custom config --> |
|
| 345 |
|
|
| 346 |
<!--delim character used to seperate fields in hdfs dump files <property> --> |
|
| 347 |
<property> |
|
| 348 |
<name>stats.delim</name> |
|
| 349 |
<value>${Stats_delim_Character}</value>
|
|
| 350 |
</property> |
|
| 351 |
<!--default string for Null String Values --> |
|
| 352 |
<property> |
|
| 353 |
<name>stats.nullString</name> |
|
| 354 |
<value>${Stats_null_String_Field}</value>
|
|
| 355 |
</property> |
|
| 356 |
<!--default string for Null Numeric Values --> |
|
| 357 |
<property> |
|
| 358 |
<name>stats.nullNum</name> |
|
| 359 |
<value>${Stats_null_Numeric_Field}</value>
|
|
| 360 |
</property> |
|
| 361 |
<!--source hbase table --> |
|
| 362 |
<property> |
|
| 363 |
<name>hbase.mapreduce.inputtable</name> |
|
| 364 |
<value>${Stats_Hbase_Source_Table}</value>
|
|
| 365 |
</property> |
|
| 366 |
<!-- This directory does not correspond to a data store. In fact, this directory only |
|
| 367 |
contains multiple data stores. It has to be set to the name of the workflow node. |
|
| 368 |
--> |
|
| 369 |
<property> |
|
| 370 |
<name>mapred.output.dir</name> |
|
| 371 |
<value>${Stats_output_Path}</value>
|
|
| 372 |
</property> |
|
| 373 |
<!-- ## Workflow node parameters --> |
|
| 374 |
<property> |
|
| 375 |
<name>mapred.reduce.tasks</name> |
|
| 376 |
<value>1</value> |
|
| 377 |
</property> |
|
| 378 |
</configuration> |
|
| 379 |
</map-reduce> |
|
| 380 |
<ok to="end" /> |
|
| 381 |
<error to="fail" /> |
|
| 382 |
</action> |
|
| 383 |
|
|
| 384 |
|
|
| 385 |
<action name="prepareDatabase"> |
|
| 386 |
<java> |
|
| 387 |
|
|
| 388 |
<prepare> |
|
| 389 |
</prepare> |
|
| 390 |
<configuration> |
|
| 391 |
<property> |
|
| 392 |
<name>mapred.job.queue.name</name> |
|
| 393 |
<value>${queueName}</value>
|
|
| 394 |
</property> |
|
| 395 |
</configuration> |
|
| 396 |
|
|
| 397 |
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class> |
|
| 398 |
<arg>-SworkingDir=${workingDir}</arg>
|
|
| 399 |
<arg>eu.dnetlib.iis.core.workflows.stats.DBInitWrapper</arg> |
|
| 400 |
|
|
| 401 |
<arg>-PStats_db_Url=${Stats_db_Url}</arg>
|
|
| 402 |
<arg>-PStats_db_User=${Stats_db_User}</arg>
|
|
| 403 |
<arg>-PStats_db_Pass=${Stats_db_Pass}</arg>
|
|
| 404 |
<arg>-PStats_db_Name=${Stats_db_Name}</arg>
|
|
| 405 |
<arg>-PStats_db_Driver=${Stats_db_Driver}</arg>
|
|
| 406 |
|
|
| 407 |
</java> |
|
| 408 |
<ok to="finalizeDatabase" /> |
|
| 409 |
<error to="fail" /> |
|
| 410 |
</action> |
|
| 411 |
|
|
| 412 |
<action name="sqoopImport"> |
|
| 413 |
<java> |
|
| 414 |
<prepare> |
|
| 415 |
</prepare> |
|
| 416 |
<configuration> |
|
| 417 |
<property> |
|
| 418 |
<name>mapred.job.queue.name</name> |
|
| 419 |
<value>${queueName}</value>
|
|
| 420 |
</property> |
|
| 421 |
</configuration> |
|
| 422 |
|
|
| 423 |
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class> |
|
| 424 |
<arg>-SworkingDir=${workingDir}</arg>
|
|
| 425 |
<arg>eu.dnetlib.iis.core.workflows.stats.SqoopWrapper</arg> |
|
| 426 |
|
|
| 427 |
<arg>-PStats_db_Url=${Stats_db_Url}</arg>
|
|
| 428 |
<arg>-PStats_db_User=${Stats_db_User}</arg>
|
|
| 429 |
<arg>-PStats_db_Pass=${Stats_db_Pass}</arg>
|
|
| 430 |
<arg>-PStats_db_Name=${Stats_db_Name}</arg>
|
|
| 431 |
<arg>-PStats_output_Path=${Stats_output_Path}</arg>
|
|
| 432 |
<arg>-PStats_sqoop_RecsPerStatement=${Stats_sqoop_RecsPerStatement}</arg>
|
|
| 433 |
<arg>-PStats_sqoop_ReducersCount=${Stats_sqoop_ReducersCount}</arg>
|
|
| 434 |
<arg>-PStats_sqoop_StatementPerTrans=${Stats_sqoop_StatementPerTrans}</arg>
|
|
| 435 |
<!-- <java-opts>-Dlog4j.configuration=log4jConfig</java-opts> --> |
|
| 436 |
</java> |
|
| 437 |
<ok to="end" /> |
|
| 438 |
<error to="fail" /> |
|
| 439 |
</action> |
|
| 440 |
|
|
| 441 |
<action name="sqoopDirectImport"> |
|
| 442 |
<sqoop xmlns="uri:oozie:sqoop-action:0.2"> |
|
| 443 |
<job-tracker>${jobTracker}</job-tracker>
|
|
| 444 |
<name-node>${nameNode}</name-node>
|
|
| 445 |
<prepare> |
|
| 446 |
|
|
| 447 |
|
|
| 448 |
</prepare> |
|
| 449 |
<configuration> |
|
| 450 |
<property> |
|
| 451 |
<name>mapred.compress.map.output</name> |
|
| 452 |
<value>false</value> |
|
| 453 |
</property> |
|
| 454 |
|
|
| 455 |
</configuration> |
|
| 456 |
<arg>--export</arg> |
|
| 457 |
<arg>-Dsqoop.export.records.per.statement</arg> |
|
| 458 |
<arg>${Stats_sqoop_RecsPerStatement}</arg>
|
|
| 459 |
|
|
| 460 |
<arg>-Dsqoop.export.statements.per.transaction</arg> |
|
| 461 |
<arg>${Stats_sqoop_StatementPerTrans}</arg>
|
|
| 462 |
|
|
| 463 |
<arg>--connect</arg> |
|
| 464 |
<arg>${Stats_db_Url}/${Stats_db_Name}</arg>
|
|
| 465 |
<arg>--table</arg> |
|
| 466 |
<arg>datasource</arg> |
|
| 467 |
|
|
| 468 |
<arg>--table</arg> |
|
| 469 |
<arg>datasource</arg> |
|
| 470 |
|
|
| 471 |
<arg>--export-dir</arg> |
|
| 472 |
<arg>${Stats_output_Path}/datasource-r-00000</arg>
|
|
| 473 |
<arg>--optionally-enclosed-by</arg> |
|
| 474 |
<arg>" "</arg> |
|
| 475 |
<arg>--input-fields-terminated-by</arg> |
|
| 476 |
<arg>${Stats_delim_Character }</arg>
|
|
| 477 |
<arg>--verbose</arg> |
|
| 478 |
<arg>--username</arg> |
|
| 479 |
<arg>${Stats_db_User}</arg>
|
|
| 480 |
<arg>--password</arg> |
|
| 481 |
<arg>${Stats_db_Pass}</arg>
|
|
| 482 |
<arg>--batch</arg> |
|
| 483 |
<arg>-m</arg> |
|
| 484 |
<arg>${Stats_sqoop_ReducersCount}</arg>
|
|
| 485 |
</sqoop> |
|
| 486 |
<ok to="end" /> |
|
| 487 |
<error to="fail" /> |
|
| 488 |
</action> |
|
| 489 |
<action name="finalizeDatabase"> |
|
| 490 |
<java> |
|
| 491 |
|
|
| 492 |
<prepare> |
|
| 493 |
</prepare> |
|
| 494 |
<configuration> |
|
| 495 |
<property> |
|
| 496 |
<name>mapred.job.queue.name</name> |
|
| 497 |
<value>${queueName}</value>
|
|
| 498 |
</property> |
|
| 499 |
</configuration> |
|
| 500 |
|
|
| 501 |
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class> |
|
| 502 |
<arg>-SworkingDir=${workingDir}</arg>
|
|
| 503 |
<arg>eu.dnetlib.iis.core.workflows.stats.DBFinalizeWrapper</arg> |
|
| 504 |
<arg>-PStats_db_Url=${Stats_db_Url}</arg>
|
|
| 505 |
<arg>-PStats_db_User=${Stats_db_User}</arg>
|
|
| 506 |
<arg>-PStats_db_Pass=${Stats_db_Pass}</arg>
|
|
| 507 |
<arg>-PStats_db_Name=${Stats_db_Name}</arg>
|
|
| 508 |
<arg>-PStats_db_Driver=${Stats_db_Driver}</arg>
|
|
| 509 |
</java> |
|
| 510 |
<ok to="end" /> |
|
| 511 |
<error to="fail" /> |
|
| 512 |
</action> |
|
| 513 |
|
|
| 514 |
|
|
| 515 |
<kill name="fail"> |
|
| 516 |
<message> |
|
| 517 |
Unfortunately, the process failed -- error message: [${wf:errorMessage(wf:lastErrorNode())}]
|
|
| 518 |
</message> |
|
| 519 |
</kill> |
|
| 520 |
<end name="end" /> |
|
| 521 |
</workflow-app> |
|
Also available in: Unified diff
testing direct sqoop import in oozie