sabeel
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Pull Requests Activity

Format workflow.xml

This commit is contained in:
Serafeim Chatzopoulos 2023-07-21 16:07:10 +03:00
parent c64e5e588f
commit cb0f3c50f6
1 changed files with 32 additions and 150 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

182
dhp-workflows/dhp-impact-indicators/src/main/resources/eu/dnetlib/dhp/oa/graph/impact_indicators/oozie_app/workflow.xml
View File

@ -13,7 +13,6 @@
</global> </global>
<!-- start using a decision node, so as to determine from which point onwards a job will continue --> <!-- start using a decision node, so as to determine from which point onwards a job will continue -->
<!-- <start to="get-doi-synonyms" /> -->
<start to="entry-point-decision" /> <start to="entry-point-decision" />
<decision name="entry-point-decision"> <decision name="entry-point-decision">
@ -38,27 +37,14 @@
</switch> </switch>
</decision> </decision>
<!-- maps openaire ids to their synonyms --> <!-- initial step: create citation network -->
<action name="create-openaire-ranking-graph"> <action name="create-openaire-ranking-graph">
<!-- This is required as a tag for spark jobs, regardless of programming language -->
<spark xmlns="uri:oozie:spark-action:0.2"> <spark xmlns="uri:oozie:spark-action:0.2">
<!-- Delete previously created doi synonym folder -->
<!-- I think we don't need this given we don't have synonyms anymore
<prepare>
<delete path="${synonymFolder}"/>
</prepare>
-->
<!-- using configs from an example on openaire -->
<master>yarn-cluster</master> <master>yarn-cluster</master>
<mode>cluster</mode> <mode>cluster</mode>
<name>OpenAIRE Ranking Graph Creation</name>
<!-- This is the name of our job -->
<name>Openaire Ranking Graph Creation</name>
<!-- Script name goes here -->
<jar>create_openaire_ranking_graph.py</jar> <jar>create_openaire_ranking_graph.py</jar>
<!-- spark configuration options: I've taken most of them from an example from dhp workflows / Master value stolen from sandro -->
<spark-opts> <spark-opts>
--executor-memory=${sparkHighExecutorMemory} --executor-memory=${sparkHighExecutorMemory}
@ -80,39 +66,30 @@
<arg>${sparkShufflePartitions}</arg> <arg>${sparkShufflePartitions}</arg>
<!-- The output of the graph should be the openaire input graph for ranking--> <!-- The output of the graph should be the openaire input graph for ranking-->
<arg>${openaireGraphInputPath}</arg> <arg>${openaireGraphInputPath}</arg>
<!-- This needs to point to the file on the hdfs i think -->
<file>${wfAppPath}/create_openaire_ranking_graph.py#create_openaire_ranking_graph.py</file> <file>${wfAppPath}/create_openaire_ranking_graph.py#create_openaire_ranking_graph.py</file>
</spark> </spark>
<!-- Do this after finishing okay -->
<ok to="non-iterative-rankings" /> <ok to="non-iterative-rankings" />
<!-- Go there if we have an error -->
<error to="openaire-graph-error" /> <error to="openaire-graph-error" />
</action> </action>
<!-- Citation Count and RAM are calculated in parallel--> <!-- Citation Count and RAM are calculated in parallel-->
<!-- Impulse Requires resources and will be run after-->
<fork name="non-iterative-rankings"> <fork name="non-iterative-rankings">
<path start="spark-cc"/> <path start="spark-cc"/>
<!-- <path start="spark-impulse"/> --> <!-- <path start="spark-impulse"/> -->
<path start="spark-ram"/> <path start="spark-ram"/>
</fork> </fork>
<!-- CC here --> <!-- Run Citation Count calculation -->
<action name="spark-cc"> <action name="spark-cc">
<!-- This is required as a tag for spark jobs, regardless of programming language -->
<spark xmlns="uri:oozie:spark-action:0.2"> <spark xmlns="uri:oozie:spark-action:0.2">
<!-- using configs from an example on openaire -->
<master>yarn-cluster</master> <master>yarn-cluster</master>
<mode>cluster</mode> <mode>cluster</mode>
<name>Citation Count calculation</name>
<!-- This is the name of our job -->
<name>Spark CC</name>
<!-- Script name goes here -->
<jar>CC.py</jar> <jar>CC.py</jar>
<!-- spark configuration options: I've taken most of them from an example from dhp workflows / Master value stolen from sandro -->
<spark-opts> <spark-opts>
--executor-memory=${sparkHighExecutorMemory} --executor-memory=${sparkHighExecutorMemory}
@ -129,31 +106,23 @@
<arg>${openaireGraphInputPath}</arg> <arg>${openaireGraphInputPath}</arg>
<!-- number of partitions to be used on joins --> <!-- number of partitions to be used on joins -->
<arg>${sparkShufflePartitions}</arg> <arg>${sparkShufflePartitions}</arg>
<!-- This needs to point to the file on the hdfs i think -->
<file>${wfAppPath}/bip-ranker/CC.py#CC.py</file> <file>${wfAppPath}/bip-ranker/CC.py#CC.py</file>
</spark> </spark>
<!-- Do this after finishing okay -->
<ok to="join-non-iterative-rankings" /> <ok to="join-non-iterative-rankings" />
<!-- Go there if we have an error -->
<error to="cc-fail" /> <error to="cc-fail" />
</action> </action>
<!-- IMPULSE here --> <!-- RAM calculation -->
<action name="spark-ram"> <action name="spark-ram">
<!-- This is required as a tag for spark jobs, regardless of programming language -->
<spark xmlns="uri:oozie:spark-action:0.2"> <spark xmlns="uri:oozie:spark-action:0.2">
<!-- using configs from an example on openaire -->
<master>yarn-cluster</master> <master>yarn-cluster</master>
<mode>cluster</mode> <mode>cluster</mode>
<name>RAM calculation</name>
<!-- This is the name of our job -->
<name>Spark RAM</name>
<!-- Script name goes here -->
<jar>TAR.py</jar> <jar>TAR.py</jar>
<!-- spark configuration options: I've taken most of them from an example from dhp workflows / Master value stolen from sandro -->
<spark-opts> <spark-opts>
--executor-memory=${sparkHighExecutorMemory} --executor-memory=${sparkHighExecutorMemory}
@ -171,37 +140,27 @@
<arg>${ramGamma}</arg> <arg>${ramGamma}</arg>
<arg>${currentYear}</arg> <arg>${currentYear}</arg>
<arg>RAM</arg> <arg>RAM</arg>
<!-- number of partitions to be used on joins -->
<arg>${sparkShufflePartitions}</arg> <arg>${sparkShufflePartitions}</arg>
<arg>${checkpointDir}</arg> <arg>${checkpointDir}</arg>
<!-- This needs to point to the file on the hdfs i think -->
<file>${wfAppPath}/bip-ranker/TAR.py#TAR.py</file> <file>${wfAppPath}/bip-ranker/TAR.py#TAR.py</file>
</spark> </spark>
<!-- Do this after finishing okay -->
<ok to="join-non-iterative-rankings" /> <ok to="join-non-iterative-rankings" />
<!-- Go there if we have an error -->
<error to="ram-fail" /> <error to="ram-fail" />
</action> </action>
<!-- JOIN NON-ITERATIVE METHODS AND THEN CONTINUE TO ITERATIVE ONES --> <!-- Join non-iterative methods -->
<join name="join-non-iterative-rankings" to="spark-impulse"/> <join name="join-non-iterative-rankings" to="spark-impulse"/>
<!-- IMPULSE here -->
<action name="spark-impulse"> <action name="spark-impulse">
<!-- This is required as a tag for spark jobs, regardless of programming language -->
<spark xmlns="uri:oozie:spark-action:0.2"> <spark xmlns="uri:oozie:spark-action:0.2">
<!-- using configs from an example on openaire -->
<master>yarn-cluster</master> <master>yarn-cluster</master>
<mode>cluster</mode> <mode>cluster</mode>
<name>Impulse calculation</name>
<!-- This is the name of our job -->
<name>Spark Impulse</name>
<!-- Script name goes here -->
<jar>CC.py</jar> <jar>CC.py</jar>
<!-- spark configuration options: I've taken most of them from an example from dhp workflows / Master value stolen from sandro -->
<spark-opts> <spark-opts>
--executor-memory=${sparkHighExecutorMemory} --executor-memory=${sparkHighExecutorMemory}
@ -219,47 +178,22 @@
<!-- number of partitions to be used on joins --> <!-- number of partitions to be used on joins -->
<arg>${sparkShufflePartitions}</arg> <arg>${sparkShufflePartitions}</arg>
<arg>3</arg> <arg>3</arg>
<!-- This needs to point to the file on the hdfs i think -->
<file>${wfAppPath}/bip-ranker/CC.py#CC.py</file> <file>${wfAppPath}/bip-ranker/CC.py#CC.py</file>
</spark> </spark>
<!-- Do this after finishing okay -->
<ok to="spark-pagerank" /> <ok to="spark-pagerank" />
<!-- Go there if we have an error -->
<error to="impulse-fail" /> <error to="impulse-fail" />
</action> </action>
<!-- Removed for ser to make pagerank & attrank serial -->
<!--
<fork name="iterative-rankings">
<path start="spark-pagerank"/>
<path start="spark-attrank"/>
</fork>
-->
<!-- PAGERANK here -->
<action name="spark-pagerank"> <action name="spark-pagerank">
<!-- This is required as a tag for spark jobs, regardless of programming language -->
<spark xmlns="uri:oozie:spark-action:0.2"> <spark xmlns="uri:oozie:spark-action:0.2">
<!-- we could add map-reduce configs here, but I don't know if we need them -->
<!-- This is the type of master-client configuration for running spark -->
<!-- <master>yarn-client</master> -->
<!-- Reference says: The master element indicates the url of the Spark Master. Ex: spark://host:port, mesos://host:port, yarn-cluster, yarn-master, or local. -->
<!-- <master>local[*]</master> -->
<!-- Reference says: The mode element if present indicates the mode of spark, where to run spark driver program. Ex: client,cluster. | In my case I always have a client -->
<!-- <mode>client</mode> -->
<!-- using configs from an example on openaire -->
<master>yarn-cluster</master> <master>yarn-cluster</master>
<mode>cluster</mode> <mode>cluster</mode>
<name>Pagerank calculation</name>
<!-- This is the name of our job -->
<name>Spark Pagerank</name>
<!-- Script name goes here -->
<jar>PageRank.py</jar> <jar>PageRank.py</jar>
<!-- spark configuration options: I've taken most of them from an example from dhp workflows / Master value stolen from sandro -->
<spark-opts> <spark-opts>
--executor-memory=${sparkHighExecutorMemory} --executor-memory=${sparkHighExecutorMemory}
@ -280,31 +214,22 @@
<!-- number of partitions to be used on joins --> <!-- number of partitions to be used on joins -->
<arg>${sparkShufflePartitions}</arg> <arg>${sparkShufflePartitions}</arg>
<arg>dfs</arg> <arg>dfs</arg>
<!-- This needs to point to the file on the hdfs i think -->
<file>${wfAppPath}/bip-ranker/PageRank.py#PageRank.py</file> <file>${wfAppPath}/bip-ranker/PageRank.py#PageRank.py</file>
</spark> </spark>
<!-- Do this after finishing okay -->
<ok to="spark-attrank" /> <ok to="spark-attrank" />
<!-- Go there if we have an error -->
<error to="pagerank-fail" /> <error to="pagerank-fail" />
</action> </action>
<!-- ATTRANK here -->
<action name="spark-attrank"> <action name="spark-attrank">
<!-- This is required as a tag for spark jobs, regardless of programming language -->
<spark xmlns="uri:oozie:spark-action:0.2"> <spark xmlns="uri:oozie:spark-action:0.2">
<!-- using configs from an example on openaire -->
<master>yarn-cluster</master> <master>yarn-cluster</master>
<mode>cluster</mode> <mode>cluster</mode>
<name>AttRank calculation</name>
<!-- This is the name of our job -->
<name>Spark AttRank</name>
<!-- Script name goes here -->
<jar>AttRank.py</jar> <jar>AttRank.py</jar>
<!-- spark configuration options: I've taken most of them from an example from dhp workflows / Master value stolen from sandro -->
<spark-opts> <spark-opts>
--executor-memory=${sparkHighExecutorMemory} --executor-memory=${sparkHighExecutorMemory}
@ -330,27 +255,16 @@
<!-- number of partitions to be used on joins --> <!-- number of partitions to be used on joins -->
<arg>${sparkShufflePartitions}</arg> <arg>${sparkShufflePartitions}</arg>
<arg>dfs</arg> <arg>dfs</arg>
<!-- This needs to point to the file on the hdfs i think -->
<file>${wfAppPath}/bip-ranker/AttRank.py#AttRank.py</file> <file>${wfAppPath}/bip-ranker/AttRank.py#AttRank.py</file>
</spark> </spark>
<!-- Do this after finishing okay -->
<ok to="get-file-names" /> <ok to="get-file-names" />
<!-- Go there if we have an error -->
<error to="attrank-fail" /> <error to="attrank-fail" />
</action> </action>
<!-- Removed for ser -->
<!--
JOIN ITERATIVE METHODS AND THEN END =
<join name="join-iterative-rankings" to="end" />
to="get-file-names"/>
-->
<!-- This will be a shell action that will output key-value pairs for output files -->
<action name="get-file-names"> <action name="get-file-names">
<!-- This is required as a tag for shell jobs -->
<shell xmlns="uri:oozie:shell-action:0.3"> <shell xmlns="uri:oozie:shell-action:0.3">
<!-- Exec is needed for shell commands - points to type of shell command --> <!-- Exec is needed for shell commands - points to type of shell command -->
@ -360,15 +274,12 @@
<!-- We only pass the directory where we expect to find the rankings --> <!-- We only pass the directory where we expect to find the rankings -->
<argument>${workingDir}</argument> <argument>${workingDir}</argument>
<!-- the name of the file run -->
<file>${wfAppPath}/get_ranking_files.sh#get_ranking_files.sh</file> <file>${wfAppPath}/get_ranking_files.sh#get_ranking_files.sh</file>
<!-- Get the output in order to be usable by following actions --> <!-- Get the output in order to be usable by following actions -->
<capture-output/> <capture-output/>
</shell> </shell>
<!-- Do this after finishing okay -->
<ok to="format-result-files" /> <ok to="format-result-files" />
<!-- Go there if we have an error -->
<error to="filename-getting-error" /> <error to="filename-getting-error" />
</action> </action>
@ -383,18 +294,12 @@
<!-- Format json files --> <!-- Format json files -->
<!-- Two parts: a) format files b) make the file endings .json.gz --> <!-- Two parts: a) format files b) make the file endings .json.gz -->
<action name="format-json-files"> <action name="format-json-files">
<!-- This is required as a tag for spark jobs, regardless of programming language -->
<spark xmlns="uri:oozie:spark-action:0.2"> <spark xmlns="uri:oozie:spark-action:0.2">
<!-- using configs from an example on openaire -->
<master>yarn-cluster</master> <master>yarn-cluster</master>
<mode>cluster</mode> <mode>cluster</mode>
<!-- This is the name of our job -->
<name>Format Ranking Results JSON</name> <name>Format Ranking Results JSON</name>
<!-- Script name goes here -->
<jar>format_ranking_results.py</jar> <jar>format_ranking_results.py</jar>
<!-- spark configuration options: I've taken most of them from an example from dhp workflows / Master value stolen from sandro -->
<spark-opts> <spark-opts>
--executor-memory=${sparkNormalExecutorMemory} --executor-memory=${sparkNormalExecutorMemory}
@ -419,13 +324,11 @@
<arg>${sparkShufflePartitions}</arg> <arg>${sparkShufflePartitions}</arg>
<!-- Type of data to be produced [bip (dois) / openaire (openaire-ids) ] --> <!-- Type of data to be produced [bip (dois) / openaire (openaire-ids) ] -->
<arg>openaire</arg> <arg>openaire</arg>
<!-- This needs to point to the file on the hdfs i think -->
<file>${wfAppPath}/format_ranking_results.py#format_ranking_results.py</file> <file>${wfAppPath}/format_ranking_results.py#format_ranking_results.py</file>
</spark> </spark>
<!-- Do this after finishing okay -->
<ok to="join-file-formatting" /> <ok to="join-file-formatting" />
<!-- Go there if we have an error -->
<error to="json-formatting-fail" /> <error to="json-formatting-fail" />
</action> </action>
@ -471,18 +374,15 @@
<file>${wfAppPath}/format_ranking_results.py#format_ranking_results.py</file> <file>${wfAppPath}/format_ranking_results.py#format_ranking_results.py</file>
</spark> </spark>
<!-- Do this after finishing okay -->
<ok to="join-file-formatting" /> <ok to="join-file-formatting" />
<!-- Go there if we have an error -->
<error to="bip-formatting-fail" /> <error to="bip-formatting-fail" />
</action> </action>
<!-- Finish formatting data and end --> <!-- Finish formatting jobs -->
<join name="join-file-formatting" to="map-openaire-to-doi"/> <join name="join-file-formatting" to="map-openaire-to-doi"/>
<!-- maps openaire ids to their synonyms --> <!-- maps openaire ids to DOIs -->
<action name="map-openaire-to-doi"> <action name="map-openaire-to-doi">
<!-- This is required as a tag for spark jobs, regardless of programming language -->
<spark xmlns="uri:oozie:spark-action:0.2"> <spark xmlns="uri:oozie:spark-action:0.2">
<!-- Delete previously created doi synonym folder --> <!-- Delete previously created doi synonym folder -->
@ -490,15 +390,10 @@
<delete path="${synonymFolder}"/> <delete path="${synonymFolder}"/>
</prepare> </prepare>
<!-- using configs from an example on openaire -->
<master>yarn-cluster</master> <master>yarn-cluster</master>
<mode>cluster</mode> <mode>cluster</mode>
<!-- This is the name of our job -->
<name>Openaire-DOI synonym collection</name> <name>Openaire-DOI synonym collection</name>
<!-- Script name goes here -->
<jar>map_openaire_ids_to_dois.py</jar> <jar>map_openaire_ids_to_dois.py</jar>
<!-- spark configuration options: I've taken most of them from an example from dhp workflows / Master value stolen from sandro -->
<spark-opts> <spark-opts>
--executor-memory=${sparkHighExecutorMemory} --executor-memory=${sparkHighExecutorMemory}
@ -515,19 +410,16 @@
<arg>${openaireDataInput}/</arg> <arg>${openaireDataInput}/</arg>
<!-- number of partitions to be used on joins --> <!-- number of partitions to be used on joins -->
<arg>${synonymFolder}</arg> <arg>${synonymFolder}</arg>
<!-- This needs to point to the file on the hdfs i think -->
<file>${wfAppPath}/map_openaire_ids_to_dois.py#map_openaire_ids_to_dois.py</file> <file>${wfAppPath}/map_openaire_ids_to_dois.py#map_openaire_ids_to_dois.py</file>
</spark> </spark>
<!-- Do this after finishing okay -->
<ok to="map-scores-to-dois" /> <ok to="map-scores-to-dois" />
<!-- Go there if we have an error -->
<error to="synonym-collection-fail" /> <error to="synonym-collection-fail" />
</action> </action>
<!-- mapping openaire scores to DOIs -->
<!-- maps openaire ids to their synonyms -->
<action name="map-scores-to-dois"> <action name="map-scores-to-dois">
<!-- This is required as a tag for spark jobs, regardless of programming language --> <!-- This is required as a tag for spark jobs, regardless of programming language -->
<spark xmlns="uri:oozie:spark-action:0.2"> <spark xmlns="uri:oozie:spark-action:0.2">
@ -535,12 +427,8 @@
<!-- using configs from an example on openaire --> <!-- using configs from an example on openaire -->
<master>yarn-cluster</master> <master>yarn-cluster</master>
<mode>cluster</mode> <mode>cluster</mode>
<!-- This is the name of our job -->
<name>Mapping Openaire Scores to DOIs</name> <name>Mapping Openaire Scores to DOIs</name>
<!-- Script name goes here -->
<jar>map_scores_to_dois.py</jar> <jar>map_scores_to_dois.py</jar>
<!-- spark configuration options: I've taken most of them from an example from dhp workflows / Master value stolen from sandro -->
<spark-opts> <spark-opts>
--executor-memory=${sparkHighExecutorMemory} --executor-memory=${sparkHighExecutorMemory}
@ -564,18 +452,15 @@
<arg>${nameNode}/${workingDir}/${wf:actionData('get-file-names')['impulse_file']}</arg> <arg>${nameNode}/${workingDir}/${wf:actionData('get-file-names')['impulse_file']}</arg>
<arg>${nameNode}/${workingDir}/${wf:actionData('get-file-names')['ram_file']}</arg> <arg>${nameNode}/${workingDir}/${wf:actionData('get-file-names')['ram_file']}</arg>
<!-- This needs to point to the file on the hdfs i think -->
<file>${wfAppPath}/map_scores_to_dois.py#map_scores_to_dois.py</file> <file>${wfAppPath}/map_scores_to_dois.py#map_scores_to_dois.py</file>
</spark> </spark>
<!-- Do this after finishing okay -->
<ok to="delete-output-path-for-actionset" /> <ok to="delete-output-path-for-actionset" />
<!-- This is the initial code <ok to="delete-output-path-for-actionset" /> -->
<!-- Go there if we have an error -->
<error to="map-scores-fail" /> <error to="map-scores-fail" />
</action> </action>
<!-- Re-create folder for result and project actionsets -->
<action name="delete-output-path-for-actionset"> <action name="delete-output-path-for-actionset">
<fs> <fs>
<delete path="${actionSetOutputPath}/results/"/> <delete path="${actionSetOutputPath}/results/"/>
@ -590,11 +475,13 @@
<action name="create-actionset-for-results"> <action name="create-actionset-for-results">
<spark xmlns="uri:oozie:spark-action:0.2"> <spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master> <master>yarn-cluster</master>
<mode>cluster</mode> <mode>cluster</mode>
<name>Produces the atomic action with the bip finder scores for publications</name> <name>Produces the atomic action with the bip finder scores for publications</name>
<class>eu.dnetlib.dhp.actionmanager.bipfinder.SparkAtomicActionScoreJob</class> <class>eu.dnetlib.dhp.actionmanager.bipfinder.SparkAtomicActionScoreJob</class>
<jar>dhp-aggregation-${projectVersion}.jar</jar> <jar>dhp-aggregation-${projectVersion}.jar</jar>
<spark-opts> <spark-opts>
--executor-memory=${sparkNormalExecutorMemory} --executor-memory=${sparkNormalExecutorMemory}
--executor-cores=${sparkExecutorCores} --executor-cores=${sparkExecutorCores}
@ -609,23 +496,19 @@
<arg>--outputPath</arg><arg>${actionSetOutputPath}/results/</arg> <arg>--outputPath</arg><arg>${actionSetOutputPath}/results/</arg>
<arg>--targetEntity</arg><arg>result</arg> <arg>--targetEntity</arg><arg>result</arg>
</spark> </spark>
<ok to="project-impact-indicators"/> <ok to="project-impact-indicators"/>
<error to="actionset-creation-fail"/> <error to="actionset-creation-fail"/>
</action> </action>
<action name="project-impact-indicators"> <action name="project-impact-indicators">
<!-- This is required as a tag for spark jobs, regardless of programming language -->
<spark xmlns="uri:oozie:spark-action:0.2"> <spark xmlns="uri:oozie:spark-action:0.2">
<!-- using configs from an example on openaire -->
<master>yarn-cluster</master> <master>yarn-cluster</master>
<mode>cluster</mode> <mode>cluster</mode>
<name>Project Impact Indicators calculation</name>
<!-- This is the name of our job -->
<name>Project Impact Indicators</name>
<!-- Script name goes here -->
<jar>projects_impact.py</jar> <jar>projects_impact.py</jar>
<!-- spark configuration options: I've taken most of them from an example from dhp workflows / Master value stolen from sandro -->
<spark-opts> <spark-opts>
--executor-memory=${sparkHighExecutorMemory} --executor-memory=${sparkHighExecutorMemory}
@ -639,7 +522,6 @@
</spark-opts> </spark-opts>
<!-- Script arguments here --> <!-- Script arguments here -->
<!-- graph data folder from which to read relations --> <!-- graph data folder from which to read relations -->
<arg>${openaireDataInput}/relation</arg> <arg>${openaireDataInput}/relation</arg>
@ -653,26 +535,23 @@
<arg>${sparkShufflePartitions}</arg> <arg>${sparkShufflePartitions}</arg>
<arg>${projectImpactIndicatorsOutput}</arg> <arg>${projectImpactIndicatorsOutput}</arg>
<!-- This needs to point to the file on the hdfs i think -->
<file>${wfAppPath}/projects_impact.py#projects_impact.py</file> <file>${wfAppPath}/projects_impact.py#projects_impact.py</file>
</spark> </spark>
<!-- Do this after finishing okay -->
<ok to="create-actionset-for-projects" /> <ok to="create-actionset-for-projects" />
<!-- Go there if we have an error -->
<error to="project-impact-indicators-fail" /> <error to="project-impact-indicators-fail" />
</action> </action>
<action name="create-actionset-for-projects"> <action name="create-actionset-for-projects">
<spark xmlns="uri:oozie:spark-action:0.2"> <spark xmlns="uri:oozie:spark-action:0.2">
<master>yarn-cluster</master> <master>yarn-cluster</master>
<mode>cluster</mode> <mode>cluster</mode>
<name>Produces the atomic action with the bip finder scores for projects</name> <name>Produces the atomic action with the bip finder scores for projects</name>
<class>eu.dnetlib.dhp.actionmanager.bipfinder.SparkAtomicActionScoreJob</class> <class>eu.dnetlib.dhp.actionmanager.bipfinder.SparkAtomicActionScoreJob</class>
<jar>dhp-aggregation-${projectVersion}.jar</jar> <jar>dhp-aggregation-${projectVersion}.jar</jar>
<spark-opts> <spark-opts>
--executor-memory=${sparkNormalExecutorMemory} --executor-memory=${sparkNormalExecutorMemory}
--executor-cores=${sparkExecutorCores} --executor-cores=${sparkExecutorCores}
@ -683,12 +562,15 @@
--conf spark.eventLog.dir=${nameNode}${spark2EventLogDir} --conf spark.eventLog.dir=${nameNode}${spark2EventLogDir}
--conf spark.sql.warehouse.dir=${sparkSqlWarehouseDir} --conf spark.sql.warehouse.dir=${sparkSqlWarehouseDir}
</spark-opts> </spark-opts>
<arg>--inputPath</arg><arg>${projectImpactIndicatorsOutput}</arg> <arg>--inputPath</arg><arg>${projectImpactIndicatorsOutput}</arg>
<arg>--outputPath</arg><arg>${actionSetOutputPath}/projects/</arg> <arg>--outputPath</arg><arg>${actionSetOutputPath}/projects/</arg>
<arg>--targetEntity</arg><arg>project</arg> <arg>--targetEntity</arg><arg>project</arg>
</spark> </spark>
<ok to="end"/> <ok to="end"/>
<error to="actionset-project-creation-fail"/> <error to="actionset-project-creation-fail"/>
</action> </action>
<!-- Definitions of failure messages --> <!-- Definitions of failure messages -->