przemyslaw.jacewicz
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Merge pull request 'fixed the result_country definition and updated the stats DB copy procedure' (#412) from antonis.lempesis/dnet-hadoop:beta into master 

Reviewed-on: D-Net/dnet-hadoop#412
This commit is contained in:
Claudio Atzori 2024-04-03 12:34:17 +02:00
parent f01390702e b7c8acc563
commit 5add51f38c
7 changed files with 90 additions and 39 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
dhp-workflows/dhp-stats-hist-snaps/src/main/resources/eu/dnetlib/dhp/oa/graph/stats-hist-snaps/oozie_app/copyDataToImpalaCluster.sh
View File

@ -9,15 +9,27 @@ fi
export HADOOP_USER_NAME=$2 export HADOOP_USER_NAME=$2
IMPALA_HDFS_NODE='' IMPALA_HDFS_NODE=''
if hdfs dfs -test -e hdfs://impala-cluster-mn1.openaire.eu >/dev/null 2>&1; then COUNTER=0
IMPALA_HDFS_NODE='hdfs://impala-cluster-mn1.openaire.eu:8020'
elif hdfs dfs -test -e hdfs://impala-cluster-mn2.openaire.eu >/dev/null 2>&1; then while [ $COUNTER -lt 3 ]; do
IMPALA_HDFS_NODE='hdfs://impala-cluster-mn2.openaire.eu:8020' if hdfs dfs -test -e hdfs://impala-cluster-mn1.openaire.eu/tmp >/dev/null 2>&1; then
else IMPALA_HDFS_NODE='hdfs://impala-cluster-mn1.openaire.eu:8020'
echo -e "\n\nPROBLEM WHEN SETTING THE HDFS-NODE FOR IMPALA CLUSTER!\n\n" break
elif hdfs dfs -test -e hdfs://impala-cluster-mn2.openaire.eu/tmp >/dev/null 2>&1; then
IMPALA_HDFS_NODE='hdfs://impala-cluster-mn2.openaire.eu:8020'
break
else
IMPALA_HDFS_NODE=''
sleep 1
fi
((COUNTER++))
done
if [ -z "$IMPALA_HDFS_NODE" ]; then
echo -e "\n\nPROBLEM WHEN SETTING THE HDFS-NODE FOR IMPALA CLUSTER! $COUNTER\n\n"
exit 1 exit 1
fi fi
echo "Active IMPALA HDFS Node: ${IMPALA_HDFS_NODE}" echo "Active IMPALA HDFS Node: ${IMPALA_HDFS_NODE} , after ${COUNTER} retries."
function copydb() { function copydb() {

27
dhp-workflows/dhp-stats-monitor-irish/src/main/resources/eu/dnetlib/dhp/oa/graph/stats-monitor-irish/oozie_app/copyDataToImpalaCluster.sh
View File

@ -9,15 +9,28 @@ fi
export HADOOP_USER_NAME=$2 export HADOOP_USER_NAME=$2
IMPALA_HDFS_NODE='' IMPALA_HDFS_NODE=''
if hdfs dfs -test -e hdfs://impala-cluster-mn1.openaire.eu >/dev/null 2>&1; then COUNTER=0
IMPALA_HDFS_NODE='hdfs://impala-cluster-mn1.openaire.eu:8020'
elif hdfs dfs -test -e hdfs://impala-cluster-mn2.openaire.eu >/dev/null 2>&1; then while [ $COUNTER -lt 3 ]; do
IMPALA_HDFS_NODE='hdfs://impala-cluster-mn2.openaire.eu:8020' if hdfs dfs -test -e hdfs://impala-cluster-mn1.openaire.eu/tmp >/dev/null 2>&1; then
else IMPALA_HDFS_NODE='hdfs://impala-cluster-mn1.openaire.eu:8020'
echo -e "\n\nPROBLEM WHEN SETTING THE HDFS-NODE FOR IMPALA CLUSTER!\n\n" break
elif hdfs dfs -test -e hdfs://impala-cluster-mn2.openaire.eu/tmp >/dev/null 2>&1; then
IMPALA_HDFS_NODE='hdfs://impala-cluster-mn2.openaire.eu:8020'
break
else
IMPALA_HDFS_NODE=''
sleep 1
fi
((COUNTER++))
done
if [ -z "$IMPALA_HDFS_NODE" ]; then
echo -e "\n\nPROBLEM WHEN SETTING THE HDFS-NODE FOR IMPALA CLUSTER! $COUNTER\n\n"
exit 1 exit 1
fi fi
echo "Active IMPALA HDFS Node: ${IMPALA_HDFS_NODE}" echo "Active IMPALA HDFS Node: ${IMPALA_HDFS_NODE} , after ${COUNTER} retries."
function copydb() { function copydb() {

27
dhp-workflows/dhp-stats-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats-monitor/oozie_app/copyDataToImpalaCluster.sh
View File

@ -9,15 +9,28 @@ fi
#export HADOOP_USER_NAME=$2 #export HADOOP_USER_NAME=$2
IMPALA_HDFS_NODE='' IMPALA_HDFS_NODE=''
if hdfs dfs -test -e hdfs://impala-cluster-mn1.openaire.eu >/dev/null 2>&1; then COUNTER=0
IMPALA_HDFS_NODE='hdfs://impala-cluster-mn1.openaire.eu:8020'
elif hdfs dfs -test -e hdfs://impala-cluster-mn2.openaire.eu >/dev/null 2>&1; then while [ $COUNTER -lt 3 ]; do
IMPALA_HDFS_NODE='hdfs://impala-cluster-mn2.openaire.eu:8020' if hdfs dfs -test -e hdfs://impala-cluster-mn1.openaire.eu/tmp >/dev/null 2>&1; then
else IMPALA_HDFS_NODE='hdfs://impala-cluster-mn1.openaire.eu:8020'
echo -e "\n\nPROBLEM WHEN SETTING THE HDFS-NODE FOR IMPALA CLUSTER!\n\n" break
elif hdfs dfs -test -e hdfs://impala-cluster-mn2.openaire.eu/tmp >/dev/null 2>&1; then
IMPALA_HDFS_NODE='hdfs://impala-cluster-mn2.openaire.eu:8020'
break
else
IMPALA_HDFS_NODE=''
sleep 1
fi
((COUNTER++))
done
if [ -z "$IMPALA_HDFS_NODE" ]; then
echo -e "\n\nPROBLEM WHEN SETTING THE HDFS-NODE FOR IMPALA CLUSTER! $COUNTER\n\n"
exit 1 exit 1
fi fi
echo "Active IMPALA HDFS Node: ${IMPALA_HDFS_NODE}" echo "Active IMPALA HDFS Node: ${IMPALA_HDFS_NODE} , after ${COUNTER} retries."
function copydb() { function copydb() {

27
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/copyDataToImpalaCluster.sh
View File

@ -7,15 +7,28 @@ then
fi fi
IMPALA_HDFS_NODE='' IMPALA_HDFS_NODE=''
if hdfs dfs -test -e hdfs://impala-cluster-mn1.openaire.eu >/dev/null 2>&1; then COUNTER=0
IMPALA_HDFS_NODE='hdfs://impala-cluster-mn1.openaire.eu:8020'
elif hdfs dfs -test -e hdfs://impala-cluster-mn2.openaire.eu >/dev/null 2>&1; then while [ $COUNTER -lt 3 ]; do
IMPALA_HDFS_NODE='hdfs://impala-cluster-mn2.openaire.eu:8020' if hdfs dfs -test -e hdfs://impala-cluster-mn1.openaire.eu/tmp >/dev/null 2>&1; then
else IMPALA_HDFS_NODE='hdfs://impala-cluster-mn1.openaire.eu:8020'
echo -e "\n\nPROBLEM WHEN SETTING THE HDFS-NODE FOR IMPALA CLUSTER!\n\n" break
elif hdfs dfs -test -e hdfs://impala-cluster-mn2.openaire.eu/tmp >/dev/null 2>&1; then
IMPALA_HDFS_NODE='hdfs://impala-cluster-mn2.openaire.eu:8020'
break
else
IMPALA_HDFS_NODE=''
sleep 1
fi
((COUNTER++))
done
if [ -z "$IMPALA_HDFS_NODE" ]; then
echo -e "\n\nPROBLEM WHEN SETTING THE HDFS-NODE FOR IMPALA CLUSTER! $COUNTER\n\n"
exit 1 exit 1
fi fi
echo "Active IMPALA HDFS Node: ${IMPALA_HDFS_NODE}" echo "Active IMPALA HDFS Node: ${IMPALA_HDFS_NODE} , after ${COUNTER} retries."
export HADOOP_USER_NAME=$6 export HADOOP_USER_NAME=$6
export PROD_USAGE_STATS_DB="openaire_prod_usage_stats" export PROD_USAGE_STATS_DB="openaire_prod_usage_stats"

4
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/monitor.sh
View File

@ -85,12 +85,12 @@ hive $HIVE_OPTS --database ${2}_funded -e "show tables" | grep -v WARN | sed "s/
hive -f foo hive -f foo
echo "Updated shadow monitor funded database" echo "Updated shadow monitor funded database"
echo "Updating shadow monitor insitutions database" echo "Updating shadow monitor institutions database"
hive -e "drop database if exists ${SHADOW}_institutions cascade" hive -e "drop database if exists ${SHADOW}_institutions cascade"
hive -e "create database if not exists ${SHADOW}_institutions" hive -e "create database if not exists ${SHADOW}_institutions"
hive $HIVE_OPTS --database ${2}_institutions -e "show tables" | grep -v WARN | sed "s/\(.*\)/create view ${SHADOW}_institutions.\1 as select * from ${2}_institutions.\1;/" > foo hive $HIVE_OPTS --database ${2}_institutions -e "show tables" | grep -v WARN | sed "s/\(.*\)/create view ${SHADOW}_institutions.\1 as select * from ${2}_institutions.\1;/" > foo
hive -f foo hive -f foo
echo "Shadow db monitor insitutions ready!" echo "Shadow db monitor institutions ready!"
echo "Updating shadow monitor RIs database" echo "Updating shadow monitor RIs database"
for i in $contexts for i in $contexts

8
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16-createIndicatorsTables.sql
View File

@ -335,8 +335,8 @@ select ar.organization, rf.no_result_fair/ar.no_allresults org_fairness
from allresults ar from allresults ar
join result_fair rf on rf.organization=ar.organization; /*EOS*/ join result_fair rf on rf.organization=ar.organization; /*EOS*/
DROP VIEW result_fair; /*EOS*/ DROP VIEW result_fair;
DROP VIEW allresults; /*EOS*/ DROP VIEW allresults;
CREATE TEMPORARY VIEW result_fair as CREATE TEMPORARY VIEW result_fair as
select year, ro.organization organization, count(distinct ro.id) no_result_fair from ${stats_db_name}.result_organization ro select year, ro.organization organization, count(distinct ro.id) no_result_fair from ${stats_db_name}.result_organization ro
@ -1000,13 +1000,13 @@ left outer join (
drop table if exists ${stats_db_name}.result_country purge; /*EOS*/ drop table if exists ${stats_db_name}.result_country purge; /*EOS*/
create table ${stats_db_name}.result_country stored as parquet as create table ${stats_db_name}.result_country stored as parquet as
select distinct ro.id, coalesce(o.country, f.country) as country select distinct ro.id, coalesce(o.country, f.country)
from ${stats_db_name}.result_organization ro from ${stats_db_name}.result_organization ro
left outer join ${stats_db_name}.organization o on o.id=ro.organization left outer join ${stats_db_name}.organization o on o.id=ro.organization
left outer join ${stats_db_name}.result_projects rp on rp.id=ro.id left outer join ${stats_db_name}.result_projects rp on rp.id=ro.id
left outer join ${stats_db_name}.project p on p.id=rp.project left outer join ${stats_db_name}.project p on p.id=rp.project
left outer join ${stats_db_name}.funder f on f.name=p.funder left outer join ${stats_db_name}.funder f on f.name=p.funder
where coalesce(o.country, f.country) IS NOT NULL; /*EOS*/ where coalesce(o.country, f.country) IS NOT NULL;
drop table if exists ${stats_db_name}.indi_result_oa_with_license purge; /*EOS*/ drop table if exists ${stats_db_name}.indi_result_oa_with_license purge; /*EOS*/
create table ${stats_db_name}.indi_result_oa_with_license stored as parquet as create table ${stats_db_name}.indi_result_oa_with_license stored as parquet as

10
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step20-createMonitorDBAll.sql
View File

@ -78,14 +78,14 @@ create table TARGET.result stored as parquet as
'openorgs____::4d4051b56708688235252f1d8fddb8c1', -- Iscte - Instituto Universitário de Lisboa 'openorgs____::4d4051b56708688235252f1d8fddb8c1', -- Iscte - Instituto Universitário de Lisboa
'openorgs____::5d55fb216b14691cf68218daf5d78cd9', -- Munster Technological University 'openorgs____::5d55fb216b14691cf68218daf5d78cd9', -- Munster Technological University
'openorgs____::0fccc7640f0cb44d5cd1b06b312a06b9', -- Cardiff University 'openorgs____::0fccc7640f0cb44d5cd1b06b312a06b9', -- Cardiff University
'openorgs____::8839b55dae0c84d56fd533f52d5d483a', -- Leibniz Institute of Ecological Urban and Regional Development 'openorgs____::8839b55dae0c84d56fd533f52d5d483a', -- Leibniz Institute of Ecological Urban and Regional Development
'openorgs____::526468206bca24c1c90da6a312295cf4', -- Cyprus University of Technology 'openorgs____::526468206bca24c1c90da6a312295cf4', -- Cyprus University of Technology
'openorgs____::b5ca9d4340e26454e367e2908ef3872f', -- Alma Mater Studiorum University of Bologna 'openorgs____::b5ca9d4340e26454e367e2908ef3872f', -- Alma Mater Studiorum University of Bologna
'openorgs____::a6340e6ecf60f6bba163659df985b0f2', -- TU Dresden 'openorgs____::a6340e6ecf60f6bba163659df985b0f2', -- TU Dresden
'openorgs____::64badd35233ba2cd4946368ef2f4cf57', -- University of Vienna 'openorgs____::64badd35233ba2cd4946368ef2f4cf57', -- University of Vienna
'openorgs____::7501d66d2297a963ebfb075c43fff88e', -- Royal Institute of Technology 'openorgs____::7501d66d2297a963ebfb075c43fff88e', -- Royal Institute of Technology
'openorgs____::d5eb679abdd31f70fcd4c8ba711148bf', -- Sorbonne University 'openorgs____::d5eb679abdd31f70fcd4c8ba711148bf', -- Sorbonne University
'openorgs____::b316f25380d106aac402f5ae8653910d' -- Centre for Research on Ecology and Forestry Applications 'openorgs____::b316f25380d106aac402f5ae8653910d' -- Centre for Research on Ecology and Forestry Applications
) )) foo; ) )) foo;
create view if not exists TARGET.category as select * from SOURCE.category; create view if not exists TARGET.category as select * from SOURCE.category;