diff --git a/dhp-workflows/dhp-monitor-update/pom.xml b/dhp-workflows/dhp-monitor-update/pom.xml deleted file mode 100644 index ca0bb9837d..0000000000 --- a/dhp-workflows/dhp-monitor-update/pom.xml +++ /dev/null @@ -1,32 +0,0 @@ - - - - dhp-workflows - eu.dnetlib.dhp - 1.2.4-SNAPSHOT - - 4.0.0 - dhp-monitor-update - - - org.apache.spark - spark-core_2.11 - - - org.apache.spark - spark-sql_2.11 - - - - - - pl.project13.maven - git-commit-id-plugin - 2.1.11 - - false - - - - - diff --git a/dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/config-default.xml b/dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/config-default.xml deleted file mode 100644 index 63fc84d75e..0000000000 --- a/dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/config-default.xml +++ /dev/null @@ -1,34 +0,0 @@ - - - jobTracker - ${jobTracker} - - - nameNode - ${nameNode} - - - oozie.use.system.libpath - true - - - oozie.action.sharelib.for.spark - spark2 - - - hive_metastore_uris - thrift://iis-cdh5-test-m3.ocean.icm.edu.pl:9083 - - - hive_jdbc_url - jdbc:hive2://iis-cdh5-test-m3.ocean.icm.edu.pl:10000/;UseNativeQuery=1;?spark.executor.memory=19166291558;spark.yarn.executor.memoryOverhead=3225;spark.driver.memory=11596411699;spark.yarn.driver.memoryOverhead=1228 - - - oozie.wf.workflow.notification.url - {serviceUrl}/v1/oozieNotification/jobUpdate?jobId=$jobId%26status=$status - - - stats_tool_api_url - ${stats_tool_api_url} - - \ No newline at end of file diff --git a/dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/monitor.sh b/dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/monitor.sh deleted file mode 100644 index 10c1ed4cad..0000000000 --- a/dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/monitor.sh +++ /dev/null @@ -1,18 +0,0 @@ -export PYTHON_EGG_CACHE=/home/$(whoami)/.python-eggs -export link_folder=/tmp/impala-shell-python-egg-cache-$(whoami) -if ! [ -L $link_folder ] -then - rm -Rf "$link_folder" - ln -sfn ${PYTHON_EGG_CACHE}${link_folder} ${link_folder} -fi - -export SOURCE=$1 -export TARGET=$2 -export SCRIPT_PATH=$3 - -echo "Getting file from " $3 -hdfs dfs -copyToLocal $3 - -echo "Updating monitor database" -cat createMonitorDB.sql | sed s/SOURCE/$1/g | sed s/TARGET/$2/g1 | impala-shell -f - -echo "Impala shell finished" diff --git a/dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/scripts/createMonitorDB.sql b/dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/scripts/createMonitorDB.sql deleted file mode 100644 index 265610e900..0000000000 --- a/dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/scripts/createMonitorDB.sql +++ /dev/null @@ -1,146 +0,0 @@ -DROP TABLE IF EXISTS TARGET.result_new; - -create table TARGET.result_new as - select distinct * from ( - select * from SOURCE.result r where exists (select 1 from SOURCE.result_organization ro where ro.id=r.id and ro.organization in ( --- 'openorgs____::b8b8ca674452579f3f593d9f5e557483', -- University College Cork --- 'openorgs____::38d7097854736583dde879d12dacafca' -- Brown University - 'openorgs____::57784c9e047e826fefdb1ef816120d92' --Arts et Métiers ParisTech - ) )) foo; - -COMPUTE STATS TARGET.result_new; - -INSERT INTO TARGET.result select * from TARGET.result_new; -COMPUTE STATS TARGET.result; - -INSERT INTO TARGET.result_citations select * from TARGET.result_citations orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_citations; - -INSERT INTO TARGET.result_references_oc select * from TARGET.result_references_oc orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_references_oc; - -INSERT INTO TARGET.result_citations_oc select * from TARGET.result_citations_oc orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_citations_oc; - -INSERT INTO TARGET.result_classifications select * from TARGET.result_classifications orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_classifications; - -INSERT INTO TARGET.result_apc select * from TARGET.result_apc orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_apc; - -INSERT INTO TARGET.result_concepts select * from TARGET.result_concepts orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_concepts; - -INSERT INTO TARGET.result_datasources select * from TARGET.result_datasources orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_datasources; - -INSERT INTO TARGET.result_fundercount select * from TARGET.result_fundercount orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_fundercount; - -INSERT INTO TARGET.result_gold select * from TARGET.result_gold orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_gold; - -INSERT INTO TARGET.result_greenoa select * from TARGET.result_greenoa orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_greenoa; - -INSERT INTO TARGET.result_languages select * from TARGET.result_languages orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_languages; - -INSERT INTO TARGET.result_licenses select * from TARGET.result_licenses orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_licenses; - -INSERT INTO TARGET.result_oids select * from TARGET.result_oids orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_oids; - -INSERT INTO TARGET.result_organization select * from TARGET.result_organization orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_organization; - -INSERT INTO TARGET.result_peerreviewed select * from TARGET.result_peerreviewed orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_peerreviewed; - -INSERT INTO TARGET.result_pids select * from TARGET.result_pids orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_pids; - -INSERT INTO TARGET.result_projectcount select * from TARGET.result_projectcount orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_projectcount; - -INSERT INTO TARGET.result_projects select * from TARGET.result_projects orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_projects; - -INSERT INTO TARGET.result_refereed select * from TARGET.result_refereed orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_refereed; - -INSERT INTO TARGET.result_sources select * from TARGET.result_sources orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_sources; - -INSERT INTO TARGET.result_topics select * from TARGET.result_topics orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_topics; - -INSERT INTO TARGET.result_fos select * from TARGET.result_fos orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.result_fos; - -create view TARGET.foo1 as select * from TARGET.result_result rr where rr.source in (select id from TARGET.result_new); -create view TARGET.foo2 as select * from TARGET.result_result rr where rr.target in (select id from TARGET.result_new); -INSERT INTO TARGET.result_result select distinct * from (select * from TARGET.foo1 union all select * from TARGET.foo2) foufou; -drop view TARGET.foo1; -drop view TARGET.foo2; -COMPUTE STATS TARGET.result_result; - --- indicators --- Sprint 1 ---- -INSERT INTO TARGET.indi_pub_green_oa select * from TARGET.indi_pub_green_oa orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.indi_pub_green_oa; -INSERT INTO TARGET.indi_pub_grey_lit select * from TARGET.indi_pub_grey_lit orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.indi_pub_grey_lit; -INSERT INTO TARGET.indi_pub_doi_from_crossref select * from TARGET.indi_pub_doi_from_crossref orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.indi_pub_doi_from_crossref; --- Sprint 2 ---- -INSERT INTO TARGET.indi_result_has_cc_licence select * from TARGET.indi_result_has_cc_licence orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.indi_result_has_cc_licence; -INSERT INTO TARGET.indi_result_has_cc_licence_url select * from TARGET.indi_result_has_cc_licence_url orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.indi_result_has_cc_licence_url; -INSERT INTO TARGET.indi_pub_has_abstract select * from TARGET.indi_pub_has_abstract orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.indi_pub_has_abstract; -INSERT INTO TARGET.indi_result_with_orcid select * from TARGET.indi_result_with_orcid orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.indi_result_with_orcid; ----- Sprint 3 ---- -INSERT INTO TARGET.indi_funded_result_with_fundref select * from TARGET.indi_funded_result_with_fundref orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.indi_funded_result_with_fundref; ----- Sprint 4 ---- -INSERT INTO TARGET.indi_pub_diamond select * from TARGET.indi_pub_diamond orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.indi_pub_diamond; -INSERT INTO TARGET.indi_pub_in_transformative select * from TARGET.indi_pub_in_transformative orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.indi_pub_in_transformative; -INSERT INTO TARGET.indi_pub_closed_other_open select * from TARGET.indi_pub_closed_other_open orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.indi_pub_closed_other_open; ----- Sprint 5 ---- -INSERT INTO TARGET.indi_result_no_of_copies select * from TARGET.indi_result_no_of_copies orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.indi_result_no_of_copies; ----- Sprint 6 ---- -INSERT INTO TARGET.indi_pub_hybrid_oa_with_cc select * from TARGET.indi_pub_hybrid_oa_with_cc orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.indi_pub_hybrid_oa_with_cc; -INSERT INTO TARGET.indi_pub_downloads select * from TARGET.indi_pub_downloads orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id); -COMPUTE STATS TARGET.indi_pub_downloads; -INSERT INTO TARGET.indi_pub_downloads_datasource select * from TARGET.indi_pub_downloads_datasource orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id); -COMPUTE STATS TARGET.indi_pub_downloads_datasource; -INSERT INTO TARGET.indi_pub_downloads_year select * from TARGET.indi_pub_downloads_year orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id); -COMPUTE STATS TARGET.indi_pub_downloads_year; -INSERT INTO TARGET.indi_pub_downloads_datasource_year select * from TARGET.indi_pub_downloads_datasource_year orig where exists (select 1 from TARGET.result_new r where r.id=orig.result_id); -COMPUTE STATS TARGET.indi_pub_downloads_datasource_year; ----- Sprint 7 ---- -INSERT INTO TARGET.indi_pub_gold_oa select * from TARGET.indi_pub_gold_oa orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.indi_pub_gold_oa; -INSERT INTO TARGET.indi_pub_hybrid select * from TARGET.indi_pub_hybrid orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.indi_pub_hybrid; - -INSERT INTO TARGET.indi_pub_has_preprint select * from TARGET.indi_pub_has_preprint orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.indi_pub_has_preprint; -INSERT INTO TARGET.indi_pub_in_subscribed select * from TARGET.indi_pub_in_subscribed orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.indi_pub_in_subscribed; -INSERT INTO TARGET.indi_result_with_pid select * from TARGET.indi_result_with_pid orig where exists (select 1 from TARGET.result_new r where r.id=orig.id); -COMPUTE STATS TARGET.indi_result_with_pid; ---create table TARGET.indi_datasets_gold_oa stored as parquet as select * from SOURCE.indi_datasets_gold_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id); ---compute stats TARGET.indi_datasets_gold_oa; ---create table TARGET.indi_software_gold_oa stored as parquet as select * from SOURCE.indi_software_gold_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id); ---compute stats TARGET.indi_software_gold_oa; -DROP TABLE TARGET.result_new; diff --git a/dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/workflow.xml b/dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/workflow.xml deleted file mode 100644 index dda645d8f4..0000000000 --- a/dhp-workflows/dhp-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/monitor/oozie_app/workflow.xml +++ /dev/null @@ -1,77 +0,0 @@ - - - - stats_db_name - the target stats database name - - - stats_db_shadow_name - the name of the shadow schema - - - monitor_db_name - the target monitor db name - - - monitor_db_shadow_name - the name of the shadow monitor db - - - stats_tool_api_url - The url of the API of the stats tool. Is used to trigger the cache update. - - - hive_metastore_uris - hive server metastore URIs - - - hive_jdbc_url - hive server jdbc url - - - hive_timeout - the time period, in seconds, after which Hive fails a transaction if a Hive client has not sent a hearbeat. The default value is 300 seconds. - - - context_api_url - the base url of the context api (https://services.openaire.eu/openaire) - - - - - ${jobTracker} - ${nameNode} - - - hive.metastore.uris - ${hive_metastore_uris} - - - hive.txn.timeout - ${hive_timeout} - - - - - - - - Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] - - - - - ${jobTracker} - ${nameNode} - monitor.sh - ${stats_db_name} - ${monitor_db_name} - ${wf:appPath()}/scripts/createMonitorDB.sql - monitor.sh - - - - - - - \ No newline at end of file