|
1
|
<workflow-app xmlns="uri:oozie:workflow:0.4" name="mainworkflows_importer_plaintext_cached_by_checksum">
|
|
2
|
|
|
3
|
<parameters>
|
|
4
|
<property>
|
|
5
|
<name>input</name>
|
|
6
|
<description>input document content directory</description>
|
|
7
|
</property>
|
|
8
|
<property>
|
|
9
|
<name>output</name>
|
|
10
|
<description>plaintext importer output directory</description>
|
|
11
|
</property>
|
|
12
|
<property>
|
|
13
|
<name>content_connection_timeout</name>
|
|
14
|
<value>60000</value>
|
|
15
|
<description>streaming content connection timeout</description>
|
|
16
|
</property>
|
|
17
|
<property>
|
|
18
|
<name>content_read_timeout</name>
|
|
19
|
<value>60000</value>
|
|
20
|
<description>streaming content read timeout</description>
|
|
21
|
</property>
|
|
22
|
<property>
|
|
23
|
<name>zk_session_timeout</name>
|
|
24
|
<value>60000</value>
|
|
25
|
<description>zookeeper session timeout when handling locks</description>
|
|
26
|
</property>
|
|
27
|
<property>
|
|
28
|
<name>default_cache_location</name>
|
|
29
|
<description>default cache location stored in HDFS</description>
|
|
30
|
</property>
|
|
31
|
<property>
|
|
32
|
<name>mapred_max_split_size</name>
|
|
33
|
<value>50000</value>
|
|
34
|
<description>maximum input data split size, required by streaming version reading DocumentContentUrl to split input data into more chunks</description>
|
|
35
|
</property>
|
|
36
|
</parameters>
|
|
37
|
|
|
38
|
<global>
|
|
39
|
<job-tracker>${jobTracker}</job-tracker>
|
|
40
|
<name-node>${nameNode}</name-node>
|
|
41
|
<configuration>
|
|
42
|
<property>
|
|
43
|
<name>mapred.job.queue.name</name>
|
|
44
|
<value>${queueName}</value>
|
|
45
|
</property>
|
|
46
|
</configuration>
|
|
47
|
</global>
|
|
48
|
|
|
49
|
<start to="preprocessing" />
|
|
50
|
|
|
51
|
<action name="preprocessing">
|
|
52
|
<sub-workflow>
|
|
53
|
<app-path>${wf:appPath()}/transformers_metadataextraction_checksum_preprocessing</app-path>
|
|
54
|
<propagate-configuration/>
|
|
55
|
<configuration>
|
|
56
|
<property>
|
|
57
|
<name>workingDir</name>
|
|
58
|
<value>${workingDir}/preprocessing/working_dir</value>
|
|
59
|
</property>
|
|
60
|
<property>
|
|
61
|
<name>input</name>
|
|
62
|
<value>${input}</value>
|
|
63
|
</property>
|
|
64
|
<property>
|
|
65
|
<name>output</name>
|
|
66
|
<value>${workingDir}/preprocessing/output</value>
|
|
67
|
</property>
|
|
68
|
</configuration>
|
|
69
|
</sub-workflow>
|
|
70
|
<ok to="mainworkflows_importer_plaintext_cached"/>
|
|
71
|
<error to="fail" />
|
|
72
|
</action>
|
|
73
|
|
|
74
|
<action name="mainworkflows_importer_plaintext_cached">
|
|
75
|
<sub-workflow>
|
|
76
|
<app-path>${wf:appPath()}/mainworkflows_importer_plaintext_cached</app-path>
|
|
77
|
<propagate-configuration/>
|
|
78
|
<configuration>
|
|
79
|
<property>
|
|
80
|
<name>workingDir</name>
|
|
81
|
<value>${workingDir}/mainworkflows_importer_plaintext_cached/working_dir</value>
|
|
82
|
</property>
|
|
83
|
<property>
|
|
84
|
<!-- checksum identified input -->
|
|
85
|
<name>input</name>
|
|
86
|
<value>${workingDir}/preprocessing/output</value>
|
|
87
|
</property>
|
|
88
|
<property>
|
|
89
|
<name>output</name>
|
|
90
|
<value>${workingDir}/mainworkflows_importer_plaintext_cached/out</value>
|
|
91
|
</property>
|
|
92
|
<!-- all the other properties are autmatically propagated-->
|
|
93
|
</configuration>
|
|
94
|
</sub-workflow>
|
|
95
|
<ok to="postprocessing-text"/>
|
|
96
|
<error to="fail" />
|
|
97
|
</action>
|
|
98
|
|
|
99
|
<action name="postprocessing-text">
|
|
100
|
<sub-workflow>
|
|
101
|
<app-path>${wf:appPath()}/transformers_metadataextraction_checksum_postprocessing_text</app-path>
|
|
102
|
<propagate-configuration/>
|
|
103
|
<configuration>
|
|
104
|
<property>
|
|
105
|
<name>workingDir</name>
|
|
106
|
<value>${workingDir}/postprocessing_text/working_dir</value>
|
|
107
|
</property>
|
|
108
|
<property>
|
|
109
|
<name>input_document_content_url</name>
|
|
110
|
<value>${input}</value>
|
|
111
|
</property>
|
|
112
|
<property>
|
|
113
|
<name>input_document_text</name>
|
|
114
|
<value>${workingDir}/mainworkflows_importer_plaintext_cached/out</value>
|
|
115
|
</property>
|
|
116
|
<property>
|
|
117
|
<name>output</name>
|
|
118
|
<value>${output}</value>
|
|
119
|
</property>
|
|
120
|
</configuration>
|
|
121
|
</sub-workflow>
|
|
122
|
<ok to="end"/>
|
|
123
|
<error to="fail" />
|
|
124
|
</action>
|
|
125
|
|
|
126
|
<kill name="fail">
|
|
127
|
<message>Unfortunately, the process failed -- error message:
|
|
128
|
[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
129
|
</kill>
|
|
130
|
<end name="end" />
|
|
131
|
</workflow-app>
|