1
|
<?xml version="1.0"?>
|
2
|
<!-- Note that documentation placed in comments in this file uses the
|
3
|
"markdown" syntax (along with its way of dividing text into sections). -->
|
4
|
<workflow-app xmlns="uri:oozie:workflow:0.3" name="test-core_examples_hadoopstreaming_cloner_without_reducer_with_explicit_schema_file">
|
5
|
<start to="data_producer" />
|
6
|
<action name="data_producer">
|
7
|
<java>
|
8
|
<job-tracker>${jobTracker}</job-tracker>
|
9
|
<name-node>${nameNode}</name-node>
|
10
|
<!-- The data generated by this node is deleted in this section -->
|
11
|
<prepare>
|
12
|
<delete path="${nameNode}${workingDir}/data_producer" />
|
13
|
<mkdir path="${nameNode}${workingDir}/data_producer" />
|
14
|
</prepare>
|
15
|
<configuration>
|
16
|
<property>
|
17
|
<name>mapred.job.queue.name</name>
|
18
|
<value>${queueName}</value>
|
19
|
</property>
|
20
|
</configuration>
|
21
|
<!-- This is simple wrapper for the Java code -->
|
22
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
23
|
<!-- The business Java code that gets to be executed -->
|
24
|
<arg>eu.dnetlib.iis.core.examples.java.SampleDataProducer</arg>
|
25
|
<!-- All input and output ports have to be bound to paths in HDFS -->
|
26
|
<arg>-Operson=${workingDir}/data_producer/person</arg>
|
27
|
<arg>-Odocument=${workingDir}/data_producer/document</arg>
|
28
|
</java>
|
29
|
<ok to="python_cloner" />
|
30
|
<error to="fail" />
|
31
|
</action>
|
32
|
<action name="python_cloner">
|
33
|
<map-reduce>
|
34
|
<job-tracker>${jobTracker}</job-tracker>
|
35
|
<name-node>${nameNode}</name-node>
|
36
|
<prepare>
|
37
|
<delete path="${nameNode}${workingDir}/python_cloner"/>
|
38
|
<mkdir path="${nameNode}${workingDir}/python_cloner"/>
|
39
|
</prepare>
|
40
|
<streaming>
|
41
|
<!-- Here, we give the relative path to the script and pass it
|
42
|
the parameters of the workflow node. The script is held
|
43
|
in a directory having the same name as the workflow node.
|
44
|
|
45
|
The parameters should be passed as **named** arguments. This
|
46
|
convention of passing them as named arguments makes the code
|
47
|
more readable/maintainable.
|
48
|
-->
|
49
|
<mapper>scripts/python_cloner/cloner.py --copies 3</mapper>
|
50
|
</streaming>
|
51
|
<configuration>
|
52
|
<!-- # Standard settings for our framework -->
|
53
|
<property>
|
54
|
<name>mapred.output.format.class</name>
|
55
|
<value>com.cloudera.science.avro.streaming.AvroAsJSONOutputFormat</value>
|
56
|
</property>
|
57
|
<property>
|
58
|
<name>mapred.input.format.class</name>
|
59
|
<value>com.cloudera.science.avro.streaming.AvroAsJSONInputFormat</value>
|
60
|
</property>
|
61
|
<!-- # Custom settings for this workflow node -->
|
62
|
<!-- We do not use any reducers, so we set their number to 0 -->
|
63
|
<property>
|
64
|
<name>mapred.reduce.tasks</name>
|
65
|
<value>0</value>
|
66
|
</property>
|
67
|
<property>
|
68
|
<name>mapred.input.dir</name>
|
69
|
<value>${workingDir}/data_producer/person</value>
|
70
|
</property>
|
71
|
<!-- Path to the input schema. This is held in the same
|
72
|
directory as the script. -->
|
73
|
<property>
|
74
|
<name>input.schema.url</name>
|
75
|
<value>${wf:appPath()}/lib/scripts/python_cloner/Person.avsc</value>
|
76
|
</property>
|
77
|
<property>
|
78
|
<name>mapred.output.dir</name>
|
79
|
<value>${workingDir}/python_cloner/output</value>
|
80
|
</property>
|
81
|
<!-- Path to the output schema. This is held in the same
|
82
|
directory as the script. -->
|
83
|
<property>
|
84
|
<name>output.schema.url</name>
|
85
|
<value>${wf:appPath()}/lib/scripts/python_cloner/Person.avsc</value>
|
86
|
</property>
|
87
|
</configuration>
|
88
|
</map-reduce>
|
89
|
<ok to="cloner"/>
|
90
|
<error to="fail"/>
|
91
|
</action>
|
92
|
<action name="cloner">
|
93
|
<java>
|
94
|
<job-tracker>${jobTracker}</job-tracker>
|
95
|
<name-node>${nameNode}</name-node>
|
96
|
<!-- The data generated by this node is deleted in this section -->
|
97
|
<prepare>
|
98
|
<delete path="${nameNode}${workingDir}/cloner" />
|
99
|
<mkdir path="${nameNode}${workingDir}/cloner" />
|
100
|
</prepare>
|
101
|
<configuration>
|
102
|
<property>
|
103
|
<name>mapred.job.queue.name</name>
|
104
|
<value>${queueName}</value>
|
105
|
</property>
|
106
|
</configuration>
|
107
|
<!-- This is simple wrapper for the Java code -->
|
108
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
109
|
<!-- The business Java code that gets to be executed -->
|
110
|
<arg>eu.dnetlib.iis.core.examples.java.PersonCloner</arg>
|
111
|
<!-- All input and output ports have to be bound to paths in HDFS -->
|
112
|
<arg>-Iperson=${workingDir}/python_cloner/output</arg>
|
113
|
<arg>-Operson=${workingDir}/cloner/person</arg>
|
114
|
</java>
|
115
|
<ok to="end" />
|
116
|
<error to="fail" />
|
117
|
</action>
|
118
|
<kill name="fail">
|
119
|
<message>Unfortunately, the process failed -- error message:
|
120
|
[${wf:errorMessage(wf:lastErrorNode())}]
|
121
|
</message>
|
122
|
</kill>
|
123
|
<end name="end"/>
|
124
|
</workflow-app>
|
125
|
|
126
|
|