D-Net
/
dnet-hadoop
14
5
Fork 11
Code Issues 14 Pull Requests 12 Projects 2 Releases 2 Wiki Activity

Stats wf executed on hive only #283

Merged
claudio.atzori merged 45 commits from antonis.lempesis/dnet-hadoop:beta into beta 2023-05-02 14:05:13 +02:00
Conversation 0 Commits 45 Files Changed 21 +28 -19
5 changed files with 28 additions and 19 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files
Showing only changes of commit d05210ba99 - Show all commits

11
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/finalizedb.sh
View File

@ -10,8 +10,11 @@ export SOURCE=$1
export SHADOW=$2
echo "Updating shadow database"
hive --database ${SOURCE} -e "show tables" | grep -v WARN | sed "s/^\(.*\)/analyze table ${SOURCE}.\1 compute statistics;/" | impala-shell -c -f -
impala-shell -q "create database if not exists ${SHADOW}"
impala-shell -d ${SHADOW} -q "show tables" --delimited | sed "s/^/drop view if exists ${SHADOW}./" | sed "s/$/;/" | impala-shell -c -f -
impala-shell -d ${SOURCE} -q "show tables" --delimited | sed "s/\(.*\)/create view ${SHADOW}.\1 as select * from ${SOURCE}.\1;/" | impala-shell -c -f -
hive --database ${SOURCE} -e "show tables" | grep -v WARN | sed "s/^\(.*\)/analyze table ${SOURCE}.\1 compute statistics;/" > foo
hive -f foo
hive -e "create database if not exists ${SHADOW}"
hive --database ${SHADOW} -e "show tables" | grep -v WARN | sed "s/^/drop view if exists ${SHADOW}./" | sed "s/$/;/" > foo
hive -f foo
hive --database ${SOURCE} -e "show tables" | grep -v WARN | sed "s/\(.*\)/create view ${SHADOW}.\1 as select * from ${SOURCE}.\1;/" > foo
hive -f foo
echo "Shadow db ready!"

6
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/indicators.sh
View File

@ -13,7 +13,7 @@ echo "Getting file from " $SCRIPT_PATH
hdfs dfs -copyToLocal $SCRIPT_PATH
echo "Creating indicators"
impala-shell -q "invalidate metadata"
impala-shell -d ${TARGET} -q "show tables" --delimited | sed "s/^\(.*\)/compute stats ${TARGET}.\1;/" | impala-shell -c -f -
cat step16-createIndicatorsTables.sql | impala-shell -d $TARGET -f -
hive --database ${TARGET} -e "show tables" | grep -v WARN | sed "s/^\(.*\)/analyze table ${TARGET}.\1 compute statistics;/" > foo
hive -f foo
hive --database ${TARGET} -f step16-createIndicatorsTables.sql
echo "Indicators created"

11
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/monitor.sh
View File

@ -15,11 +15,14 @@ echo "Getting file from " $4
hdfs dfs -copyToLocal $4
echo "Creating monitor database"
cat step20-createMonitorDB.sql | sed s/SOURCE/$1/g | sed s/TARGET/$2/g1 | impala-shell -f -
cat step20-createMonitorDB.sql | sed s/SOURCE/$1/g | sed s/TARGET/$2/g1 > foo
hive -f foo
echo "Impala shell finished"
echo "Updating shadow monitor database"
impala-shell -q "create database if not exists ${SHADOW}"
impala-shell -d ${SHADOW} -q "show tables" --delimited | sed "s/^/drop view if exists ${SHADOW}./" | sed "s/$/;/" | impala-shell -f -
impala-shell -d ${TARGET} -q "show tables" --delimited | sed "s/\(.*\)/create view ${SHADOW}.\1 as select * from ${TARGET}.\1;/" | impala-shell -f -
hive -e "create database if not exists ${SHADOW}"
hive --database ${SHADOW} -e "show tables" | grep -v WARN | sed "s/^/drop view if exists ${SHADOW}./" | sed "s/$/;/" > foo
hive -f foo
hive --database ${TARGET} -e "show tables" | grep -v WARN | sed "s/\(.*\)/create view ${SHADOW}.\1 as select * from ${TARGET}.\1;/" > foo
hive -f foo
echo "Shadow db ready!"

12
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/observatory-post.sh
View File

@ -10,12 +10,14 @@ export SOURCE=$1
export TARGET=$2
export SHADOW=$3
impala-shell -q "invalidate metadata;"
impala-shell -d ${TARGET} -q "show tables" --delimited | sed "s/\(.*\)/compute stats ${TARGET}.\1;/" | impala-shell -f -
hive --database ${TARGET} -e "show tables" | grep -v WARN | sed "s/\(.*\)/analyze table ${TARGET}.\1 compute statistics;/" > foo
hive -f foo
echo "Impala shell finished"
echo "Updating shadow observatory database"
impala-shell -q "create database if not exists ${SHADOW}"
impala-shell -d ${SHADOW} -q "show tables" --delimited | sed "s/^/drop view if exists ${SHADOW}./" | sed "s/$/;/" | impala-shell -f -
impala-shell -d ${TARGET} -q "show tables" --delimited | sed "s/\(.*\)/create view ${SHADOW}.\1 as select * from ${TARGET}.\1;/" | impala-shell -f -
hive -e "create database if not exists ${SHADOW}"
hive --database ${SHADOW} -e "show tables" | grep -v WARN | sed "s/^/drop view if exists ${SHADOW}./" | sed "s/$/;/" > foo
hive -f foo
hive -d ${TARGET} -e "show tables" | grep -v WARN | sed "s/\(.*\)/create view ${SHADOW}.\1 as select * from ${TARGET}.\1;/" > foo
hive -f foo
echo "Shadow db ready!"

7
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/observatory-pre.sh
View File

@ -11,6 +11,7 @@ export TARGET=$2
export SHADOW=$3
echo "Creating observatory database"
impala-shell -q "drop database if exists ${TARGET} cascade"
impala-shell -q "create database if not exists ${TARGET}"
impala-shell -d ${SOURCE} -q "show tables" --delimited | grep -iv roar | sed "s/\(.*\)/create view ${TARGET}.\1 as select * from ${SOURCE}.\1;/" | impala-shell -f -
hive -e "drop database if exists ${TARGET} cascade"
hive -e "create database if not exists ${TARGET}"
hive --database ${SOURCE} -e "show tables" | grep -v WARN | grep -iv roar | sed "s/\(.*\)/create view ${TARGET}.\1 as select * from ${SOURCE}.\1;/" > foo
hive -f foo