1
|
<workflow-app xmlns="uri:oozie:workflow:0.3" name="test-mainworkflows_integration_import">
|
2
|
<start to="importer" />
|
3
|
<action name="importer">
|
4
|
<sub-workflow>
|
5
|
<app-path>${wf:appPath()}/mapred_import</app-path>
|
6
|
<propagate-configuration/>
|
7
|
<configuration>
|
8
|
<property>
|
9
|
<!-- TODO possibly not needed -->
|
10
|
<name>actionmanager_avro_storage_dir</name>
|
11
|
<value>${workingDir}/exporter</value>
|
12
|
</property>
|
13
|
<property>
|
14
|
<name>mapreduce_inputformat_class</name>
|
15
|
<value>eu.dnetlib.iis.mainworkflows.importer.mapred.PredefinedTableInputFormat</value>
|
16
|
</property>
|
17
|
<property>
|
18
|
<name>inputformat_idx_props_location</name>
|
19
|
<value>eu/dnetlib/iis/mainworkflows/integration/import/input/meta/mappings.properties</value>
|
20
|
</property>
|
21
|
<property>
|
22
|
<name>inference_provenance_blacklist</name>
|
23
|
<value>iis::.*</value>
|
24
|
</property>
|
25
|
<property>
|
26
|
<name>skip_deleted_by_inference</name>
|
27
|
<value>true</value>
|
28
|
</property>
|
29
|
<property>
|
30
|
<name>hbase_input_table</name>
|
31
|
<value>undefined</value>
|
32
|
</property>
|
33
|
<property>
|
34
|
<name>output</name>
|
35
|
<value>${workingDir}/mr_import</value>
|
36
|
</property>
|
37
|
</configuration>
|
38
|
</sub-workflow>
|
39
|
|
40
|
<ok to="forking" />
|
41
|
|
42
|
<error to="fail" />
|
43
|
</action>
|
44
|
|
45
|
<fork name="forking">
|
46
|
<path start="consumer-person"/>
|
47
|
<path start="consumer-project"/>
|
48
|
<path start="consumer-document"/>
|
49
|
</fork>
|
50
|
|
51
|
<action name="consumer-person">
|
52
|
<java>
|
53
|
<job-tracker>${jobTracker}</job-tracker>
|
54
|
<name-node>${nameNode}</name-node>
|
55
|
<!-- The data generated by this node is deleted in this section -->
|
56
|
<prepare>
|
57
|
<delete path="${nameNode}${workingDir}/consumer-person" />
|
58
|
<mkdir path="${nameNode}${workingDir}/consumer-person" />
|
59
|
</prepare>
|
60
|
<configuration>
|
61
|
<property>
|
62
|
<name>mapred.job.queue.name</name>
|
63
|
<value>${queueName}</value>
|
64
|
</property>
|
65
|
</configuration>
|
66
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
67
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.TestingConsumer</arg>
|
68
|
<arg>-C{person,
|
69
|
eu.dnetlib.iis.importer.schemas.Person,
|
70
|
eu/dnetlib/iis/mainworkflows/integration/import/output/person.json}</arg>
|
71
|
<arg>-SworkingDir=${workingDir}/consumer-person/working_dir</arg>
|
72
|
<arg>-Iperson=${workingDir}/mr_import/person</arg>
|
73
|
</java>
|
74
|
<ok to="joining" />
|
75
|
<error to="fail" />
|
76
|
</action>
|
77
|
|
78
|
<action name="consumer-project">
|
79
|
<java>
|
80
|
<job-tracker>${jobTracker}</job-tracker>
|
81
|
<name-node>${nameNode}</name-node>
|
82
|
<!-- The data generated by this node is deleted in this section -->
|
83
|
<prepare>
|
84
|
<delete path="${nameNode}${workingDir}/consumer-project" />
|
85
|
<mkdir path="${nameNode}${workingDir}/consumer-project" />
|
86
|
</prepare>
|
87
|
<configuration>
|
88
|
<property>
|
89
|
<name>mapred.job.queue.name</name>
|
90
|
<value>${queueName}</value>
|
91
|
</property>
|
92
|
</configuration>
|
93
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
94
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.TestingConsumer</arg>
|
95
|
<arg>-C{project,
|
96
|
eu.dnetlib.iis.importer.schemas.Project,
|
97
|
eu/dnetlib/iis/mainworkflows/integration/import/output/project.json}</arg>
|
98
|
<arg>-SworkingDir=${workingDir}/consumer-project/working_dir</arg>
|
99
|
<arg>-Iproject=${workingDir}/mr_import/project</arg>
|
100
|
</java>
|
101
|
<ok to="joining" />
|
102
|
<error to="fail" />
|
103
|
</action>
|
104
|
|
105
|
<action name="consumer-document">
|
106
|
<java>
|
107
|
<job-tracker>${jobTracker}</job-tracker>
|
108
|
<name-node>${nameNode}</name-node>
|
109
|
<!-- The data generated by this node is deleted in this section -->
|
110
|
<prepare>
|
111
|
<delete path="${nameNode}${workingDir}/consumer-document" />
|
112
|
<mkdir path="${nameNode}${workingDir}/consumer-document" />
|
113
|
</prepare>
|
114
|
<configuration>
|
115
|
<property>
|
116
|
<name>mapred.job.queue.name</name>
|
117
|
<value>${queueName}</value>
|
118
|
</property>
|
119
|
</configuration>
|
120
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
121
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.TestingConsumer</arg>
|
122
|
<arg>-C{document,
|
123
|
eu.dnetlib.iis.importer.schemas.DocumentMetadata,
|
124
|
eu/dnetlib/iis/mainworkflows/integration/import/output/document.json}</arg>
|
125
|
<arg>-SworkingDir=${workingDir}/consumer-document/working_dir</arg>
|
126
|
<arg>-Idocument=${workingDir}/mr_import/docmeta</arg>
|
127
|
</java>
|
128
|
<ok to="joining" />
|
129
|
<error to="fail" />
|
130
|
</action>
|
131
|
|
132
|
<join name="joining" to="end"/>
|
133
|
|
134
|
<kill name="fail">
|
135
|
<message>Unfortunately, the process failed -- error message:
|
136
|
[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
137
|
</kill>
|
138
|
<end name="end" />
|
139
|
</workflow-app>
|