1
|
<workflow-app xmlns="uri:oozie:workflow:0.3"
|
2
|
name="test-core_examples_javamapreduce_cloner_with_multiple_output_without_reducer_with_explicit_schema">
|
3
|
<!-- This example writes to 2 datastores: person and documents.
|
4
|
The class responsible for writing multiple datastores is:
|
5
|
eu.dnetlib.iis.core.examples.javamapreduce.PersonClonerMapperMultipleOutput. -->
|
6
|
<start to="data_producer" />
|
7
|
<action name="data_producer">
|
8
|
<java>
|
9
|
<job-tracker>${jobTracker}</job-tracker>
|
10
|
<name-node>${nameNode}</name-node>
|
11
|
<!-- The data generated by this node is deleted in this section -->
|
12
|
<prepare>
|
13
|
<delete path="${nameNode}${workingDir}/data_producer" />
|
14
|
<mkdir path="${nameNode}${workingDir}/data_producer" />
|
15
|
</prepare>
|
16
|
<configuration>
|
17
|
<property>
|
18
|
<name>mapred.job.queue.name</name>
|
19
|
<value>${queueName}</value>
|
20
|
</property>
|
21
|
</configuration>
|
22
|
<!-- This is simple wrapper for the Java code -->
|
23
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
24
|
<!-- The business Java code that gets to be executed -->
|
25
|
<arg>eu.dnetlib.iis.core.examples.java.SampleDataProducer</arg>
|
26
|
<!-- All input and output ports have to be bound to paths in HDFS -->
|
27
|
<arg>-Operson=${workingDir}/data_producer/person</arg>
|
28
|
<arg>-Odocument=${workingDir}/data_producer/document</arg>
|
29
|
</java>
|
30
|
<ok to="mr_cloner" />
|
31
|
<error to="fail" />
|
32
|
</action>
|
33
|
<action name="mr_cloner">
|
34
|
<map-reduce>
|
35
|
<job-tracker>${jobTracker}</job-tracker>
|
36
|
<name-node>${nameNode}</name-node>
|
37
|
<!-- The data generated by this node in the previous run is
|
38
|
deleted in this section -->
|
39
|
<prepare>
|
40
|
<delete path="${nameNode}${workingDir}/mr_cloner" />
|
41
|
</prepare>
|
42
|
<!-- That's a multiple output MapReduce job, so no need to
|
43
|
create mr_cloner directory, since it will be created by
|
44
|
MapReduce /> -->
|
45
|
<configuration>
|
46
|
|
47
|
<!-- # Standard set of options that stays the same regardless
|
48
|
of a concrete definition of map-reduce job -->
|
49
|
|
50
|
<!-- ## Various options -->
|
51
|
|
52
|
<!--This property seems to not be needed -->
|
53
|
<!--<property> <name>mapred.job.queue.name</name> <value>${queueName}</value>
|
54
|
</property> -->
|
55
|
<property>
|
56
|
<name>mapreduce.inputformat.class</name>
|
57
|
<value>org.apache.avro.mapreduce.AvroKeyInputFormat</value>
|
58
|
</property>
|
59
|
<!-- The output format is not needed since there is no Reduce phase -->
|
60
|
<!-- <property>
|
61
|
<name>mapreduce.outputformat.class</name>
|
62
|
<value>eu.dnetlib.iis.core.javamapreduce.hack.KeyOutputFormat</value>
|
63
|
</property>-->
|
64
|
<property>
|
65
|
<name>mapred.mapoutput.key.class</name>
|
66
|
<value>org.apache.avro.mapred.AvroKey</value>
|
67
|
</property>
|
68
|
<property>
|
69
|
<name>mapred.mapoutput.value.class</name>
|
70
|
<value>org.apache.avro.mapred.AvroValue</value>
|
71
|
</property>
|
72
|
<property>
|
73
|
<name>mapred.output.key.class</name>
|
74
|
<value>org.apache.avro.mapred.AvroKey</value>
|
75
|
</property>
|
76
|
<property>
|
77
|
<name>mapred.output.value.class</name>
|
78
|
<value>org.apache.avro.mapred.AvroValue</value>
|
79
|
</property>
|
80
|
<property>
|
81
|
<name>mapred.output.key.comparator.class</name>
|
82
|
<value>org.apache.avro.hadoop.io.AvroKeyComparator</value>
|
83
|
</property>
|
84
|
<property>
|
85
|
<name>io.serializations</name>
|
86
|
<value>org.apache.hadoop.io.serializer.WritableSerialization,org.apache.hadoop.io.serializer.avro.AvroSpecificSerialization,org.apache.hadoop.io.serializer.avro.AvroReflectSerialization,org.apache.avro.hadoop.io.AvroSerialization
|
87
|
</value>
|
88
|
</property>
|
89
|
<property>
|
90
|
<name>mapred.output.value.groupfn.class</name>
|
91
|
<value>org.apache.avro.hadoop.io.AvroKeyComparator</value>
|
92
|
</property>
|
93
|
<property>
|
94
|
<name>rpc.engine.org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolPB
|
95
|
</name>
|
96
|
<value>org.apache.hadoop.ipc.ProtobufRpcEngine</value>
|
97
|
</property> <!-- This directory does not correspond to a data store. In fact,
|
98
|
this directory only contains multiple data stores.-->
|
99
|
|
100
|
<!-- ## This is required for new MapReduce API usage -->
|
101
|
|
102
|
<property>
|
103
|
<name>mapred.mapper.new-api</name>
|
104
|
<value>true</value>
|
105
|
</property>
|
106
|
<property>
|
107
|
<name>mapred.reducer.new-api</name>
|
108
|
<value>true</value>
|
109
|
</property>
|
110
|
|
111
|
<!-- # Job-specific options -->
|
112
|
|
113
|
<!-- Since there is no reduce phase, there should be no
|
114
|
reduce tasks -->
|
115
|
<property>
|
116
|
<name>mapred.reduce.tasks</name>
|
117
|
<value>0</value>
|
118
|
</property>
|
119
|
|
120
|
<!-- ## Names of all output ports -->
|
121
|
|
122
|
<property>
|
123
|
<name>avro.mapreduce.multipleoutputs</name>
|
124
|
<value>person age</value>
|
125
|
</property>
|
126
|
|
127
|
<!-- ## Output classes for all output ports -->
|
128
|
|
129
|
<property>
|
130
|
<name>avro.mapreduce.multipleoutputs.namedOutput.person.format
|
131
|
</name>
|
132
|
<value>org.apache.avro.mapreduce.AvroKeyOutputFormat</value>
|
133
|
</property>
|
134
|
<property>
|
135
|
<name>avro.mapreduce.multipleoutputs.namedOutput.age.format
|
136
|
</name>
|
137
|
<value>org.apache.avro.mapreduce.AvroKeyOutputFormat</value>
|
138
|
</property>
|
139
|
|
140
|
<!-- ## Classes of mapper and reducer -->
|
141
|
|
142
|
<property>
|
143
|
<name>mapreduce.map.class</name>
|
144
|
<value>eu.dnetlib.iis.core.examples.javamapreduce.MultipleOutputPersonClonerMapper</value>
|
145
|
</property>
|
146
|
|
147
|
<!-- No reducer -->
|
148
|
|
149
|
<!-- ## Schemas -->
|
150
|
|
151
|
<!-- ### Shema of the data ingested by the mapper. To be more precise,
|
152
|
it's the schema of Avro data passed as template parameter of the AvroKey
|
153
|
object passed to mapper. -->
|
154
|
<property>
|
155
|
<name>avro.schema.input.key</name>
|
156
|
<value>{
|
157
|
"type" : <!-- This directory does not correspond to a data store. In fact,
|
158
|
this directory only contains multiple data stores.--> "record",
|
159
|
"name" : "Person",
|
160
|
"namespace" : "eu.dnetlib.iis.core.examples.schemas.documentandauthor",
|
161
|
"fields" : [ {
|
162
|
"name" : "id",
|
163
|
"type" : "int"
|
164
|
}, {
|
165
|
"name" : "name",
|
166
|
"type" : "string"
|
167
|
}, {
|
168
|
"name" : "age",
|
169
|
"type" : "int"
|
170
|
} ]
|
171
|
}
|
172
|
</value>
|
173
|
</property>
|
174
|
|
175
|
<!-- ### Schemas of the data produced by the mapper -->
|
176
|
|
177
|
<!-- #### Schema of the key produced by the mapper. To be more precise,
|
178
|
it's the schema of Avro data produced by the mapper and passed forward as
|
179
|
template paramter of AvroKey object. (it has to have the same value for the
|
180
|
"*.reader.schema" and "*.writer.schema"). -->
|
181
|
|
182
|
<!-- As a convention, we're setting "null" values
|
183
|
since mapper does not produce any standard data in this example
|
184
|
(probably any other valid Avro schema would do as well).-->
|
185
|
|
186
|
<property>
|
187
|
<name>avro.serialization.key.reader.schema</name>
|
188
|
<value>"null"</value>
|
189
|
</property>
|
190
|
<property>
|
191
|
<name>avro.serialization.key.writer.schema</name>
|
192
|
<value>"null"</value>
|
193
|
</property>
|
194
|
|
195
|
|
196
|
<!-- #### Schema of the value produced by the mapper. To be more precise,
|
197
|
it's the schema of Avro data produced by the mapper and passed forward as
|
198
|
template paramter of AvroValue object. (it has to have the same value for
|
199
|
the "*.reader.schema" and "*.writer.schema") -->
|
200
|
|
201
|
<!-- As a convention, we're setting "null" values
|
202
|
since mapper does not produce any standard data in this example
|
203
|
(probably any other valid Avro schema would do as well).-->
|
204
|
|
205
|
<property>
|
206
|
<name>avro.serialization.value.reader.schema</name>
|
207
|
<value>"null"</value>
|
208
|
</property>
|
209
|
<property>
|
210
|
<name>avro.serialization.value.writer.schema</name>
|
211
|
<value>"null"</value>
|
212
|
</property>
|
213
|
|
214
|
<!-- ### Shema of multiple output ports. -->
|
215
|
|
216
|
<property>
|
217
|
<name>avro.mapreduce.multipleoutputs.namedOutput.person.keyschema
|
218
|
</name>
|
219
|
<value>{
|
220
|
"type" : "record",
|
221
|
"name" : "Person",
|
222
|
"namespace" : "eu.dnetlib.iis.core.examples.schemas.documentandauthor",
|
223
|
"fields" : [ {
|
224
|
"name" : "id",
|
225
|
"type" : "int"
|
226
|
}, {
|
227
|
"name" : "name",
|
228
|
"type" : "string"
|
229
|
}, {
|
230
|
"name" : "age",
|
231
|
"type" : "int"
|
232
|
} ]
|
233
|
}
|
234
|
</value>
|
235
|
</property>
|
236
|
|
237
|
<property>
|
238
|
<name>avro.mapreduce.multipleoutputs.namedOutput.age.keyschema
|
239
|
</name>
|
240
|
<value>{
|
241
|
"type" : "record",
|
242
|
"name" : "PersonAge",
|
243
|
"namespace" : "eu.dnetlib.iis.core.examples.schemas.documentandauthor",
|
244
|
"fields" : [ {
|
245
|
"name" : "age",
|
246
|
"type" : "int"
|
247
|
} ]
|
248
|
}
|
249
|
</value>
|
250
|
</property>
|
251
|
|
252
|
<!-- ## Specification of the input and output data store -->
|
253
|
|
254
|
<property>
|
255
|
<name>mapred.input.dir</name>
|
256
|
<value>${workingDir}/data_producer/person</value>
|
257
|
</property>
|
258
|
<!-- This directory does not correspond to a data store. In fact,
|
259
|
this directory only contains multiple data stores. It has to
|
260
|
be set to the name of the workflow node.-->
|
261
|
<property>
|
262
|
<name>mapred.output.dir</name>
|
263
|
<value>${workingDir}/mr_cloner</value>
|
264
|
</property>
|
265
|
|
266
|
<!-- ## Workflow node parameters -->
|
267
|
|
268
|
<property>
|
269
|
<name>copiesCount</name>
|
270
|
<value>2</value>
|
271
|
</property>
|
272
|
</configuration>
|
273
|
</map-reduce>
|
274
|
<ok to="cloner" />
|
275
|
<error to="fail" />
|
276
|
</action>
|
277
|
|
278
|
<!-- cloner works on duplicated data -->
|
279
|
<action name="cloner">
|
280
|
<java>
|
281
|
<job-tracker>${jobTracker}</job-tracker>
|
282
|
<name-node>${nameNode}</name-node>
|
283
|
<!-- The data generated by this node is deleted in this section -->
|
284
|
<prepare>
|
285
|
<delete path="${nameNode}${workingDir}/cloner" />
|
286
|
<mkdir path="${nameNode}${workingDir}/cloner" />
|
287
|
</prepare>
|
288
|
<configuration>
|
289
|
<property>
|
290
|
<name>mapred.job.queue.name</name>
|
291
|
<value>${queueName}</value>
|
292
|
</property>
|
293
|
</configuration>
|
294
|
<!-- This is simple wrapper for the Java code -->
|
295
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
296
|
<!-- The business Java code that gets to be executed -->
|
297
|
<arg>eu.dnetlib.iis.core.examples.java.PersonCloner</arg>
|
298
|
<!-- All input and output ports have to be bound to paths in HDFS -->
|
299
|
<arg>-Iperson=${workingDir}/mr_cloner/person</arg>
|
300
|
<arg>-Operson=${workingDir}/cloner/person</arg>
|
301
|
</java>
|
302
|
<ok to="end" />
|
303
|
<error to="fail" />
|
304
|
</action>
|
305
|
<kill name="fail">
|
306
|
<message>Unfortunately, the process failed -- error message:
|
307
|
[${wf:errorMessage(wf:lastErrorNode())}]
|
308
|
</message>
|
309
|
</kill>
|
310
|
<end name="end" />
|
311
|
</workflow-app>
|