1
|
<workflow-app xmlns="uri:oozie:workflow:0.3" name="line-by-line-copier">
|
2
|
<start to="producer"/>
|
3
|
<action name="producer">
|
4
|
<java>
|
5
|
<job-tracker>${jobTracker}</job-tracker>
|
6
|
<name-node>${nameNode}</name-node>
|
7
|
<!-- The data generated by this node is deleted in this section -->
|
8
|
<prepare>
|
9
|
<delete path="${nameNode}${workingDir}/producer" />
|
10
|
<mkdir path="${nameNode}${workingDir}/producer" />
|
11
|
</prepare>
|
12
|
<configuration>
|
13
|
<property>
|
14
|
<name>mapred.job.queue.name</name>
|
15
|
<value>${queueName}</value>
|
16
|
</property>
|
17
|
</configuration>
|
18
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
19
|
<arg>eu.dnetlib.iis.core.examples.protobuf.java.TextDataProducer</arg>
|
20
|
<!-- All input and output ports have to be bound to paths in
|
21
|
HDFS, working directory has to be specified as well -->
|
22
|
<arg>-Operson=${workingDir}/producer/person.csv</arg>
|
23
|
<arg>-Odocument=${workingDir}/producer/doc.csv</arg>
|
24
|
</java>
|
25
|
<ok to="copier"/>
|
26
|
<error to="fail"/>
|
27
|
</action>
|
28
|
<action name="copier">
|
29
|
<java>
|
30
|
<job-tracker>${jobTracker}</job-tracker>
|
31
|
<name-node>${nameNode}</name-node>
|
32
|
<!-- The data generated by this node is deleted in this section -->
|
33
|
<prepare>
|
34
|
<delete path="${nameNode}${workingDir}/copier" />
|
35
|
<mkdir path="${nameNode}${workingDir}/copier" />
|
36
|
</prepare>
|
37
|
<configuration>
|
38
|
<property>
|
39
|
<name>mapred.job.queue.name</name>
|
40
|
<value>${queueName}</value>
|
41
|
</property>
|
42
|
</configuration>
|
43
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
44
|
<arg>eu.dnetlib.iis.core.examples.protobuf.java.LineByLineCopier</arg>
|
45
|
<!-- All input and output ports have to be bound to paths in
|
46
|
HDFS, working directory has to be specified as well -->
|
47
|
<arg>-Iperson=${workingDir}/producer/person.csv</arg>
|
48
|
<arg>-Idocument=${workingDir}/producer/doc.csv</arg>
|
49
|
<arg>-Operson_copy=${workingDir}/copier/person_copy.csv</arg>
|
50
|
<arg>-Odocument_copy=${workingDir}/copier/doc_copy.csv</arg>
|
51
|
</java>
|
52
|
<ok to="end"/>
|
53
|
<error to="fail"/>
|
54
|
</action>
|
55
|
<kill name="fail">
|
56
|
<message>Unfortunately, the process failed -- error message: [${wf:errorMessage(wf:lastErrorNode())}]</message>
|
57
|
</kill>
|
58
|
<end name="end"/>
|
59
|
</workflow-app>
|