1
|
<workflow-app xmlns="uri:oozie:workflow:0.4" name="mainworkflows_preprocessing_quick">
|
2
|
<parameters>
|
3
|
<!-- intermediate datastores related -->
|
4
|
<property>
|
5
|
<name>input_document_to_dataset</name>
|
6
|
<value>$UNDEFINED$</value>
|
7
|
<description>intermediate avro datastore holding document to dataset relations</description>
|
8
|
</property>
|
9
|
<property>
|
10
|
<name>input_document_to_project</name>
|
11
|
<value>$UNDEFINED$</value>
|
12
|
<description>intermediate avro datastore holding document to project relations</description>
|
13
|
</property>
|
14
|
<!-- export datacite & wos related -->
|
15
|
<property>
|
16
|
<name>import_mdstore_service_location</name>
|
17
|
<description>MDStore service (not WSDL) location URL</description>
|
18
|
</property>
|
19
|
<property>
|
20
|
<name>import_dataset_mdstore_ids_csv</name>
|
21
|
<description>dataset MDStore identifier</description>
|
22
|
</property>
|
23
|
<property>
|
24
|
<name>import_wos_mdstore_id</name>
|
25
|
<description>WoS MDStore identifier</description>
|
26
|
</property>
|
27
|
<!-- export related -->
|
28
|
<property>
|
29
|
<name>export_action_hbase_table_name</name>
|
30
|
<description>action manager hbase table name</description>
|
31
|
</property>
|
32
|
<property>
|
33
|
<name>export_action_hbase_table_initialize</name>
|
34
|
<description>flag indicating input table should be initialized</description>
|
35
|
</property>
|
36
|
<!-- action set id properties -->
|
37
|
<property>
|
38
|
<name>export_action_set_id</name>
|
39
|
<value>$UNDEFINED$</value>
|
40
|
<description>action-set identifier of exported data</description>
|
41
|
</property>
|
42
|
<property>
|
43
|
<name>export_action_set_id_document_referencedProjects</name>
|
44
|
<value>$UNDEFINED$</value>
|
45
|
<description>document_referencedProjects action-set identifier of exported data</description>
|
46
|
</property>
|
47
|
<property>
|
48
|
<name>export_action_set_id_document_referencedDatasets</name>
|
49
|
<value>$UNDEFINED$</value>
|
50
|
<description>document_referencedDatasets action-set identifier of exported data</description>
|
51
|
</property>
|
52
|
<property>
|
53
|
<name>export_action_set_id_entity_wos</name>
|
54
|
<description>action-set identifier of exported data containing wos entities</description>
|
55
|
</property>
|
56
|
<property>
|
57
|
<name>export_action_set_id_entity_dataset</name>
|
58
|
<description>action-set identifier of exported data containing dataset entities</description>
|
59
|
</property>
|
60
|
<property>
|
61
|
<name>export_action_hbase_remote_zookeeper_quorum</name>
|
62
|
<value>$UNDEFINED$</value>
|
63
|
<description>external hbase zookeeper quorum, set to empty value by default which means data will be exported to local hbase instance</description>
|
64
|
</property>
|
65
|
<property>
|
66
|
<name>export_action_hbase_remote_zookeeper_clientport</name>
|
67
|
<value>$UNDEFINED$</value>
|
68
|
<description>external hbase zookeeper client port, required only whe zookeeper quorum property is set</description>
|
69
|
</property>
|
70
|
</parameters>
|
71
|
|
72
|
<global>
|
73
|
<job-tracker>${jobTracker}</job-tracker>
|
74
|
<name-node>${nameNode}</name-node>
|
75
|
<configuration>
|
76
|
<property>
|
77
|
<name>mapred.job.queue.name</name>
|
78
|
<value>${queueName}</value>
|
79
|
</property>
|
80
|
</configuration>
|
81
|
</global>
|
82
|
|
83
|
<start to="empty_dataset_existing_id_producer" />
|
84
|
|
85
|
<!-- this node is required due to the PIG limitation
|
86
|
disallowing empty directories as input avro storages -->
|
87
|
<action name="empty_dataset_existing_id_producer">
|
88
|
<java>
|
89
|
<!-- The data generated by this node is deleted in this section -->
|
90
|
<prepare>
|
91
|
<delete path="${nameNode}${workingDir}/producer" />
|
92
|
<mkdir path="${nameNode}${workingDir}/producer" />
|
93
|
</prepare>
|
94
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
95
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.Producer</arg>
|
96
|
<arg>-C{dataset_existing_id,
|
97
|
eu.dnetlib.iis.common.schemas.DocumentId,
|
98
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
99
|
<arg>-SworkingDir=${workingDir}/producer/working_dir</arg>
|
100
|
<arg>-Odataset_existing_id=${workingDir}/producer/dataset_existing_id</arg>
|
101
|
</java>
|
102
|
<ok to="export"/>
|
103
|
<error to="fail"/>
|
104
|
</action>
|
105
|
|
106
|
<action name="export">
|
107
|
<sub-workflow>
|
108
|
<app-path>${wf:appPath()}/mainworkflows_common_export</app-path>
|
109
|
<propagate-configuration/>
|
110
|
<configuration>
|
111
|
<property>
|
112
|
<name>workingDir</name>
|
113
|
<value>${workingDir}/mainworkflows_common_export/working_dir</value>
|
114
|
</property>
|
115
|
<!-- input ports -->
|
116
|
<property>
|
117
|
<name>input_document_to_project</name>
|
118
|
<value>${input_document_to_project}</value>
|
119
|
</property>
|
120
|
<property>
|
121
|
<name>input_document_to_dataset</name>
|
122
|
<value>${input_document_to_dataset}</value>
|
123
|
</property>
|
124
|
<property>
|
125
|
<name>input_aux_dataset_existing_id</name>
|
126
|
<value>${workingDir}/producer/dataset_existing_id</value>
|
127
|
</property>
|
128
|
<!-- entities exporting modes -->
|
129
|
<property>
|
130
|
<name>active_export_referenceddataset_datasets</name>
|
131
|
<value>true</value>
|
132
|
</property>
|
133
|
<property>
|
134
|
<name>active_export_referencedproject_entities</name>
|
135
|
<value>true</value>
|
136
|
</property>
|
137
|
<property>
|
138
|
<name>mdstore_service_location</name>
|
139
|
<value>${import_mdstore_service_location}</value>
|
140
|
</property>
|
141
|
<property>
|
142
|
<name>dataset_mdstore_ids_csv</name>
|
143
|
<value>${import_dataset_mdstore_ids_csv}</value>
|
144
|
</property>
|
145
|
<property>
|
146
|
<name>wos_mdstore_id</name>
|
147
|
<value>${import_wos_mdstore_id}</value>
|
148
|
</property>
|
149
|
<!-- export related -->
|
150
|
<property>
|
151
|
<name>action_hbase_table_name</name>
|
152
|
<value>${export_action_hbase_table_name}</value>
|
153
|
</property>
|
154
|
<property>
|
155
|
<name>action_hbase_table_initialize</name>
|
156
|
<value>${export_action_hbase_table_initialize}</value>
|
157
|
</property>
|
158
|
<!-- action set id properties -->
|
159
|
<property>
|
160
|
<name>action_set_id</name>
|
161
|
<value>${export_action_set_id}</value>
|
162
|
</property>
|
163
|
<property>
|
164
|
<name>action_set_id_document_referencedProjects</name>
|
165
|
<value>${export_action_set_id_document_referencedProjects}</value>
|
166
|
</property>
|
167
|
<property>
|
168
|
<name>action_set_id_document_referencedDatasets</name>
|
169
|
<value>${export_action_set_id_document_referencedDatasets}</value>
|
170
|
</property>
|
171
|
<property>
|
172
|
<name>action_set_id_entity_wos</name>
|
173
|
<value>${export_action_set_id_entity_wos}</value>
|
174
|
</property>
|
175
|
<property>
|
176
|
<name>action_set_id_entity_dataset</name>
|
177
|
<value>${export_action_set_id_entity_dataset}</value>
|
178
|
</property>
|
179
|
<property>
|
180
|
<name>action_hbase_remote_zookeeper_quorum</name>
|
181
|
<value>${export_action_hbase_remote_zookeeper_quorum}</value>
|
182
|
</property>
|
183
|
<property>
|
184
|
<name>action_hbase_remote_zookeeper_clientport</name>
|
185
|
<value>${export_action_hbase_remote_zookeeper_clientport}</value>
|
186
|
</property>
|
187
|
</configuration>
|
188
|
</sub-workflow>
|
189
|
<ok to="end"/>
|
190
|
<error to="fail" />
|
191
|
</action>
|
192
|
|
193
|
<kill name="fail">
|
194
|
<message>Unfortunately, the process failed -- error message:
|
195
|
[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
196
|
</kill>
|
197
|
<end name="end" />
|
198
|
</workflow-app>
|