forked from D-Net/dnet-hadoop
workflow works in parallel on 2 activity files
This commit is contained in:
parent
941e94af06
commit
a1861b9eaa
|
@ -25,11 +25,7 @@ public class OrcidAuthorsDOIsDataGen extends OrcidDSManager {
|
|||
String tarGzUri =
|
||||
hdfsServerUri.concat(hdfsOrcidDefaultPath).concat(activitiesFileNameTarGz);
|
||||
Path outputPath =
|
||||
new Path(
|
||||
hdfsServerUri
|
||||
.concat(hdfsOrcidDefaultPath)
|
||||
.concat(outputAuthorsDOIsPath)
|
||||
.concat("authors_dois.seq"));
|
||||
new Path(hdfsServerUri.concat(hdfsOrcidDefaultPath).concat(outputAuthorsDOIsPath));
|
||||
ActivitiesDecompressor.parseGzActivities(conf, tarGzUri, outputPath);
|
||||
}
|
||||
|
||||
|
|
|
@ -18,11 +18,16 @@
|
|||
<delete path='${workingPath_activities}/output'/>
|
||||
<mkdir path='${workingPath_activities}/output'/>
|
||||
</fs>
|
||||
<ok to="GenerateOrcidAuthorsDOIsData"/>
|
||||
<ok to="fork_generate_orcid_authors_dois"/>
|
||||
<error to="Kill"/>
|
||||
</action>
|
||||
|
||||
<action name="GenerateOrcidAuthorsDOIsData">
|
||||
<fork name = "fork_generate_orcid_authors_dois">
|
||||
<path start = "Generate_Orcid_Authors_DOIs_0"/>
|
||||
<path start = "Generate_Orcid_Authors_DOIs_1"/>
|
||||
</fork>
|
||||
|
||||
<action name="Generate_Orcid_Authors_DOIs_0">
|
||||
<java>
|
||||
<job-tracker>${jobTracker}</job-tracker>
|
||||
<name-node>${nameNode}</name-node>
|
||||
|
@ -30,10 +35,27 @@
|
|||
<arg>-d</arg><arg>${workingPath_activities}/</arg>
|
||||
<arg>-n</arg><arg>${nameNode}</arg>
|
||||
<arg>-f</arg><arg>ORCID_2019_activites_0.tar.gz</arg>
|
||||
<arg>-o</arg><arg>output/</arg>
|
||||
<arg>-o</arg><arg>output/authors_dois_0.seq</arg>
|
||||
</java>
|
||||
<ok to="End"/>
|
||||
<ok to="join_node"/>
|
||||
<error to="Kill"/>
|
||||
</action>
|
||||
<end name="End"/>
|
||||
|
||||
<action name="Generate_Orcid_Authors_DOIs_1">
|
||||
<java>
|
||||
<job-tracker>${jobTracker}</job-tracker>
|
||||
<name-node>${nameNode}</name-node>
|
||||
<main-class>eu.dnetlib.doiboost.orcid.OrcidAuthorsDOIsDataGen</main-class>
|
||||
<arg>-d</arg><arg>${workingPath_activities}/</arg>
|
||||
<arg>-n</arg><arg>${nameNode}</arg>
|
||||
<arg>-f</arg><arg>ORCID_2019_activites_1.tar.gz</arg>
|
||||
<arg>-o</arg><arg>output/authors_dois_1.seq</arg>
|
||||
</java>
|
||||
<ok to="join_node"/>
|
||||
<error to="Kill"/>
|
||||
</action>
|
||||
|
||||
<join name = "join_node" to = "End"/>
|
||||
|
||||
<end name="End"/>
|
||||
</workflow-app>
|
Loading…
Reference in New Issue