1 |
51263
|
claudio.at
|
<workflow-app xmlns="uri:oozie:workflow:0.4" name="importer_mdrecord">
|
2 |
|
|
|
3 |
|
|
<parameters>
|
4 |
|
|
<property>
|
5 |
|
|
<name>mdstore_facade_factory_classname</name>
|
6 |
|
|
<value>eu.dnetlib.dhp.wf.importer.facade.WebServiceMDStoreFacadeFactory</value>
|
7 |
|
|
<description>ServiceFacadeFactory implementation class name producing eu.dnetlib.dhp.wf.importer.facade.WebServiceMDStoreFacade</description>
|
8 |
|
|
</property>
|
9 |
|
|
<property>
|
10 |
|
|
<name>mdstore_service_location</name>
|
11 |
|
|
<value>$UNDEFINED$</value>
|
12 |
|
|
<description>mdstore service location</description>
|
13 |
|
|
</property>
|
14 |
|
|
<property>
|
15 |
|
|
<name>mdstore_ids_csv</name>
|
16 |
|
|
<value>$UNDEFINED$</value>
|
17 |
|
|
<description>comma separated mdstore identifiers</description>
|
18 |
|
|
</property>
|
19 |
|
|
<property>
|
20 |
|
|
<name>mdstore_record_maxlength</name>
|
21 |
|
|
<value>500000</value>
|
22 |
|
|
<description>maximum allowed length of mdstore record</description>
|
23 |
|
|
</property>
|
24 |
|
|
<property>
|
25 |
|
|
<name>output</name>
|
26 |
|
|
<description>ImportedRecord avro datastore output holding mdrecords</description>
|
27 |
|
|
</property>
|
28 |
|
|
<property>
|
29 |
|
|
<name>output_report_root_path</name>
|
30 |
|
|
<description>base directory for storing reports</description>
|
31 |
|
|
</property>
|
32 |
|
|
<property>
|
33 |
|
|
<name>output_report_relative_path</name>
|
34 |
|
|
<value>import_mdrecord</value>
|
35 |
|
|
<description>directory for storing report (relative to output_report_root_path)</description>
|
36 |
|
|
</property>
|
37 |
|
|
<property>
|
38 |
|
|
<name>dnet_service_client_read_timeout</name>
|
39 |
|
|
<value>60000</value>
|
40 |
|
|
<description>DNet service client reading timeout (expressed in milliseconds)</description>
|
41 |
|
|
</property>
|
42 |
|
|
<property>
|
43 |
|
|
<name>dnet_service_client_connection_timeout</name>
|
44 |
|
|
<value>60000</value>
|
45 |
|
|
<description>DNet service client connection timeout (expressed in milliseconds)</description>
|
46 |
|
|
</property>
|
47 |
|
|
<property>
|
48 |
|
|
<name>resultset_client_read_timeout</name>
|
49 |
|
|
<value>60000</value>
|
50 |
|
|
<description>result set client reading timeout (expressed in milliseconds)</description>
|
51 |
|
|
</property>
|
52 |
|
|
<property>
|
53 |
|
|
<name>resultset_client_connection_timeout</name>
|
54 |
|
|
<value>60000</value>
|
55 |
|
|
<description>result set client connection timeout (expressed in milliseconds)</description>
|
56 |
|
|
</property>
|
57 |
|
|
<property>
|
58 |
|
|
<name>report_properties_prefix</name>
|
59 |
|
|
<value>import.mdrecord</value>
|
60 |
|
|
<description>report entry related to total number of imported records</description>
|
61 |
|
|
</property>
|
62 |
|
|
</parameters>
|
63 |
|
|
|
64 |
|
|
<global>
|
65 |
|
|
<job-tracker>${jobTracker}</job-tracker>
|
66 |
|
|
<name-node>${nameNode}</name-node>
|
67 |
|
|
<configuration>
|
68 |
|
|
<property>
|
69 |
|
|
<name>mapreduce.job.queuename</name>
|
70 |
|
|
<value>${queueName}</value>
|
71 |
|
|
</property>
|
72 |
|
|
<property>
|
73 |
|
|
<name>oozie.launcher.mapred.job.queue.name</name>
|
74 |
|
|
<value>${oozieLauncherQueueName}</value>
|
75 |
|
|
</property>
|
76 |
|
|
</configuration>
|
77 |
|
|
</global>
|
78 |
|
|
|
79 |
|
|
<start to="mdrecord-importer" />
|
80 |
|
|
|
81 |
|
|
<action name="mdrecord-importer">
|
82 |
|
|
<java>
|
83 |
|
|
<!-- The data generated by this node is deleted in this section -->
|
84 |
|
|
<prepare>
|
85 |
|
|
<delete path="${nameNode}${output}" />
|
86 |
|
|
</prepare>
|
87 |
|
|
<main-class>eu.dnetlib.dhp.common.java.ProcessWrapper</main-class>
|
88 |
|
|
<arg>eu.dnetlib.dhp.wf.importer.mdrecord.MDStoreRecordsImporter</arg>
|
89 |
|
|
|
90 |
|
|
<arg>-Pimport.mdstore.service.location=${mdstore_service_location}</arg>
|
91 |
|
|
<arg>-Pimport.mdstore.ids.csv=${mdstore_ids_csv}</arg>
|
92 |
|
|
<arg>-Pimport.mdstore.record.maxlength=${mdstore_record_maxlength}</arg>
|
93 |
|
|
|
94 |
|
|
<arg>-Pimport.resultset.client.read.timeout=${resultset_client_read_timeout}</arg>
|
95 |
|
|
<arg>-Pimport.resultset.client.connection.timeout=${resultset_client_connection_timeout}</arg>
|
96 |
|
|
<arg>-Pdnet.service.client.read.timeout=${dnet_service_client_read_timeout}</arg>
|
97 |
|
|
<arg>-Pdnet.service.client.connection.timeout=${dnet_service_client_connection_timeout}</arg>
|
98 |
|
|
|
99 |
|
|
<arg>-Pimport.facade.factory.classname=${mdstore_facade_factory_classname}</arg>
|
100 |
|
|
<arg>-Omdrecords=${output}</arg>
|
101 |
|
|
<capture-output />
|
102 |
|
|
</java>
|
103 |
|
|
<ok to="report" />
|
104 |
|
|
<error to="fail" />
|
105 |
|
|
</action>
|
106 |
|
|
|
107 |
|
|
<action name="report">
|
108 |
|
|
<java>
|
109 |
|
|
<main-class>eu.dnetlib.dhp.common.java.ProcessWrapper</main-class>
|
110 |
|
|
<arg>eu.dnetlib.dhp.common.report.ReportGenerator</arg>
|
111 |
|
|
<arg>-Preport.${report_properties_prefix}.total=${wf:actionData('mdrecord-importer')['TOTAL']}</arg>
|
112 |
|
|
<arg>-Preport.${report_properties_prefix}.invalid.sizeExceeded=${wf:actionData('mdrecord-importer')['SIZE_EXCEEDED']}</arg>
|
113 |
|
|
<arg>-Oreport=${output_report_root_path}/${output_report_relative_path}</arg>
|
114 |
|
|
</java>
|
115 |
|
|
<ok to="end" />
|
116 |
|
|
<error to="fail" />
|
117 |
|
|
</action>
|
118 |
|
|
|
119 |
|
|
<kill name="fail">
|
120 |
|
|
<message>Unfortunately, the process failed -- error message:
|
121 |
|
|
[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
122 |
|
|
</kill>
|
123 |
|
|
<end name="end" />
|
124 |
|
|
</workflow-app>
|