antonis.lempesis
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
1
Fork 1
Code Issues Pull Requests Projects Releases Wiki Activity

convert_hive_to_spark_actions #1

Merged
antonis.lempesis merged 20 commits from convert_hive_to_spark_actions into beta 2024-09-23 13:53:29 +02:00
Conversation 0 Commits 20 Files Changed 30 +190 -120
13 changed files with 190 additions and 120 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files
Showing only changes of commit e0ac494859 - Show all commits

67
dhp-workflows/dhp-stats-hist-snaps/src/main/resources/eu/dnetlib/dhp/oa/graph/stats-hist-snaps/oozie_app/copyDataToImpalaCluster.sh
View File

@ -8,6 +8,8 @@ fi
export HADOOP_USER_NAME=$2
SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR=0
# Set the active HDFS node of OCEAN and IMPALA cluster.
OCEAN_HDFS_NODE='hdfs://nameservice1'
@ -30,7 +32,9 @@ while [ $COUNTER -lt 3 ]; do
done
if [ -z "$IMPALA_HDFS_NODE" ]; then
echo -e "\n\nERROR: PROBLEM WHEN SETTING THE HDFS-NODE FOR IMPALA CLUSTER! | AFTER ${COUNTER} RETRIES.\n\n"
exit 1
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 1
fi
fi
echo -e "Active IMPALA HDFS Node: ${IMPALA_HDFS_NODE} , after ${COUNTER} retries.\n\n"
@ -39,26 +43,25 @@ IMPALA_CONFIG_FILE='/etc/impala_cluster/hdfs-site.xml'
IMPALA_HDFS_DB_BASE_PATH="${IMPALA_HDFS_NODE}/user/hive/warehouse"
# Set sed arguments.
LOCATION_HDFS_NODE_SED_ARG="s|${OCEAN_HDFS_NODE}|${IMPALA_HDFS_NODE}|g" # This requires to be used with "sed -e" in order to have the "|" delimiter (as the "/" conflicts with the URIs)
# Set the SED command arguments for column-names with reserved words:
DATE_SED_ARG_1='s/[[:space:]]\date[[:space:]]/\`date\`/g'
DATE_SED_ARG_2='s/\.date,/\.\`date\`,/g' # the "date" may be part of a larger field name like "datestamp" or "date_aggregated", so we need to be careful with what we are replacing.
DATE_SED_ARG_3='s/\.date[[:space:]]/\.\`date\` /g'
HASH_SED_ARG_1='s/[[:space:]]\hash[[:space:]]/\`hash\`/g'
HASH_SED_ARG_2='s/\.hash,/\.\`hash\`,/g'
HASH_SED_ARG_3='s/\.hash[[:space:]]/\.\`hash\` /g'
LOCATION_SED_ARG_1='s/[[:space:]]\location[[:space:]]/\`location\`/g'
LOCATION_SED_ARG_2='s/\.location,/\.\`location\`,/g'
LOCATION_SED_ARG_3='s/\.location[[:space:]]/\.\`location\` /g'
function print_elapsed_time()
{
start_time=$1
end_time=$(date +%s)
elapsed_time=$(($end_time-$start_time))
hours=$((elapsed_time / 3600))
minutes=$(((elapsed_time % 3600) / 60))
seconds=$((elapsed_time % 60))
printf "\nElapsed time: %02d:%02d:%02d\n\n" $hours $minutes $seconds
}
function copydb() {
db=$1
start_db_time=$(date +%s)
echo -e "\nStart processing db: '${db}'..\n"
# Delete the old DB from Impala cluster (if exists).
@ -67,7 +70,9 @@ function copydb() {
if [ -n "$log_errors" ]; then
echo -e "\n\nERROR: THERE WAS A PROBLEM WHEN DROPPING THE OLD DATABASE! EXITING...\n\n"
rm -f error.log
exit 2
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 2
fi
fi
echo -e "\n\nCopying files of '${db}', from Ocean to Impala cluster..\n"
@ -91,7 +96,9 @@ function copydb() {
else
echo -e "\n\nERROR: FAILED TO TRANSFER THE FILES OF '${db}', WITH 'hadoop distcp'. GOT EXIT STATUS: $?\n\n"
rm -f error.log
exit 3
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 3
fi
fi
# In case we ever use this script for a writable DB (using inserts/updates), we should perform the following costly operation as well..
@ -109,17 +116,13 @@ function copydb() {
num_tables=0
entities_on_ocean=`hive -e "show tables in ${db};" | sed 's/WARN:.*//g'` # Get the tables and views without any potential the "WARN" logs.
for i in ${entities_on_ocean[@]}; do # Use un-quoted values, as the elemetns are single-words.
for i in ${entities_on_ocean[@]}; do # Use un-quoted values, as the elements are single-words.
# Check if this is a view by showing the create-statement where it should print "create view" for a view, not the "create table". Unfortunately, there is no "show views" command.
create_entity_statement=`hive -e "show create table ${db}.${i};"` # It needs to happen in two stages, otherwise the "grep" is not able to match multi-line statement.
create_view_statement_test=`echo -e "$create_entity_statement" | grep 'CREATE VIEW'`
create_entity_statement=`hive --database ${db} -e "show create table ${i};"` # We need to use the "--database", instead of including it inside the query, in order to return the statements with the '`' chars being in the right place to be used by impala-shell. However, we need to add the db-name in the "CREATE VIEW view_name" statement.
create_view_statement_test=`echo -e "$create_entity_statement" | grep 'CREATE VIEW'` # It needs to happen in two stages, otherwise the "grep" is not able to match multi-line statement.
if [ -n "$create_view_statement_test" ]; then
echo -e "\n'${i}' is a view, so we will save its 'create view' statement and execute it on Impala, after all tables have been created.\n"
create_view_statement=`echo -e "$create_entity_statement" | sed 's/WARN:.*//g' | sed 's/\`//g' \
| sed 's/"$/;/' | sed 's/^"//' | sed 's/\\"\\"/\"/g' | sed -e "${LOCATION_HDFS_NODE_SED_ARG}" | sed "${DATE_SED_ARG_1}" | sed "${HASH_SED_ARG_1}" | sed "${LOCATION_SED_ARG_1}" \
| sed "${DATE_SED_ARG_2}" | sed "${HASH_SED_ARG_2}" | sed "${LOCATION_SED_ARG_2}" \
| sed "${DATE_SED_ARG_3}" | sed "${HASH_SED_ARG_3}" | sed "${LOCATION_SED_ARG_3}"`
create_view_statement=`echo -e "$create_entity_statement" | sed 's/WARN:.*//g' | sed 's/"$/;/' | sed 's/^"//' | sed 's/\\"\\"/\"/g' | sed -e "${LOCATION_HDFS_NODE_SED_ARG}" | sed "s/CREATE VIEW /CREATE VIEW ${db}./"`
all_create_view_statements+=("$create_view_statement")
else
echo -e "\n'${i}' is a table, so we will check for its parquet files and create the table on Impala cluster.\n"
@ -127,12 +130,17 @@ function copydb() {
CURRENT_PRQ_FILE=`hdfs dfs -conf ${IMPALA_CONFIG_FILE} -ls -C "${IMPALA_HDFS_DB_BASE_PATH}/${db}.db/${i}/" | grep -v 'Found' | grep -v '_impala_insert_staging' | head -1`
if [ -z "$CURRENT_PRQ_FILE" ]; then # If there is not parquet-file inside.
echo -e "\nERROR: THE TABLE \"${i}\" HAD NO FILES TO GET THE SCHEMA FROM! IT'S EMPTY!\n\n"
exit 4 # Comment out when testing a DB which has such a table, just for performing this exact test-check.
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 4
fi
else
impala-shell --user ${HADOOP_USER_NAME} -i ${IMPALA_HOSTNAME} -q "create table ${db}.${i} like parquet '${CURRENT_PRQ_FILE}' stored as parquet;" |& tee error.log
log_errors=`cat error.log | grep -E "WARN|ERROR|FAILED"`
if [ -n "$log_errors" ]; then
echo -e "\n\nERROR: THERE WAS A PROBLEM WHEN CREATING TABLE '${i}'!\n\n"
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 5
fi
fi
fi
fi
@ -176,7 +184,9 @@ function copydb() {
if [[ $new_num_of_views_to_retry -eq $previous_num_of_views_to_retry ]]; then
echo -e "\n\nERROR: THE NUMBER OF VIEWS TO RETRY HAS NOT BEEN REDUCED! THE SCRIPT IS LIKELY GOING TO AN INFINITE-LOOP! EXITING..\n\n"
exit 5
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 6
fi
elif [[ $new_num_of_views_to_retry -gt 0 ]]; then
echo -e "\nTo be retried \"create_view_statements\" (${new_num_of_views_to_retry}):\n\n${all_create_view_statements[@]}\n"
else
@ -204,11 +214,14 @@ function copydb() {
else
echo -e "\n\nERROR: 1 OR MORE ENTITIES OF DB '${db}' FAILED TO BE COPIED TO IMPALA CLUSTER!\n\n"
rm -f error.log
exit 6
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 7
fi
fi
rm -f error.log
echo -e "\n\nFinished processing db: ${db}\n\n"
echo -e "\n\nFinished processing db: ${db}\n"
print_elapsed_time start_db_time
}

68
dhp-workflows/dhp-stats-monitor-irish/src/main/resources/eu/dnetlib/dhp/oa/graph/stats-monitor-irish/oozie_app/copyDataToImpalaCluster.sh
View File

@ -8,6 +8,9 @@ fi
export HADOOP_USER_NAME=$2
SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR=0
# Set the active HDFS node of OCEAN and IMPALA cluster.
OCEAN_HDFS_NODE='hdfs://nameservice1'
echo -e "\nOCEAN HDFS virtual-name which resolves automatically to the active-node: ${OCEAN_HDFS_NODE}"
@ -29,7 +32,9 @@ while [ $COUNTER -lt 3 ]; do
done
if [ -z "$IMPALA_HDFS_NODE" ]; then
echo -e "\n\nERROR: PROBLEM WHEN SETTING THE HDFS-NODE FOR IMPALA CLUSTER! | AFTER ${COUNTER} RETRIES.\n\n"
exit 1
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 1
fi
fi
echo -e "Active IMPALA HDFS Node: ${IMPALA_HDFS_NODE} , after ${COUNTER} retries.\n\n"
@ -38,26 +43,25 @@ IMPALA_CONFIG_FILE='/etc/impala_cluster/hdfs-site.xml'
IMPALA_HDFS_DB_BASE_PATH="${IMPALA_HDFS_NODE}/user/hive/warehouse"
# Set sed arguments.
LOCATION_HDFS_NODE_SED_ARG="s|${OCEAN_HDFS_NODE}|${IMPALA_HDFS_NODE}|g" # This requires to be used with "sed -e" in order to have the "|" delimiter (as the "/" conflicts with the URIs)
# Set the SED command arguments for column-names with reserved words:
DATE_SED_ARG_1='s/[[:space:]]\date[[:space:]]/\`date\`/g'
DATE_SED_ARG_2='s/\.date,/\.\`date\`,/g' # the "date" may be part of a larger field name like "datestamp" or "date_aggregated", so we need to be careful with what we are replacing.
DATE_SED_ARG_3='s/\.date[[:space:]]/\.\`date\` /g'
HASH_SED_ARG_1='s/[[:space:]]\hash[[:space:]]/\`hash\`/g'
HASH_SED_ARG_2='s/\.hash,/\.\`hash\`,/g'
HASH_SED_ARG_3='s/\.hash[[:space:]]/\.\`hash\` /g'
LOCATION_SED_ARG_1='s/[[:space:]]\location[[:space:]]/\`location\`/g'
LOCATION_SED_ARG_2='s/\.location,/\.\`location\`,/g'
LOCATION_SED_ARG_3='s/\.location[[:space:]]/\.\`location\` /g'
function print_elapsed_time()
{
start_time=$1
end_time=$(date +%s)
elapsed_time=$(($end_time-$start_time))
hours=$((elapsed_time / 3600))
minutes=$(((elapsed_time % 3600) / 60))
seconds=$((elapsed_time % 60))
printf "\nElapsed time: %02d:%02d:%02d\n\n" $hours $minutes $seconds
}
function copydb() {
db=$1
start_db_time=$(date +%s)
echo -e "\nStart processing db: '${db}'..\n"
# Delete the old DB from Impala cluster (if exists).
@ -66,7 +70,9 @@ function copydb() {
if [ -n "$log_errors" ]; then
echo -e "\n\nERROR: THERE WAS A PROBLEM WHEN DROPPING THE OLD DATABASE! EXITING...\n\n"
rm -f error.log
exit 2
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 2
fi
fi
echo -e "\n\nCopying files of '${db}', from Ocean to Impala cluster..\n"
@ -90,7 +96,9 @@ function copydb() {
else
echo -e "\n\nERROR: FAILED TO TRANSFER THE FILES OF '${db}', WITH 'hadoop distcp'. GOT EXIT STATUS: $?\n\n"
rm -f error.log
exit 3
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 3
fi
fi
# In case we ever use this script for a writable DB (using inserts/updates), we should perform the following costly operation as well..
@ -108,17 +116,13 @@ function copydb() {
num_tables=0
entities_on_ocean=`hive -e "show tables in ${db};" | sed 's/WARN:.*//g'` # Get the tables and views without any potential the "WARN" logs.
for i in ${entities_on_ocean[@]}; do # Use un-quoted values, as the elemetns are single-words.
for i in ${entities_on_ocean[@]}; do # Use un-quoted values, as the elements are single-words.
# Check if this is a view by showing the create-statement where it should print "create view" for a view, not the "create table". Unfortunately, there is no "show views" command.
create_entity_statement=`hive -e "show create table ${db}.${i};"` # It needs to happen in two stages, otherwise the "grep" is not able to match multi-line statement.
create_view_statement_test=`echo -e "$create_entity_statement" | grep 'CREATE VIEW'`
create_entity_statement=`hive --database ${db} -e "show create table ${i};"` # We need to use the "--database", instead of including it inside the query, in order to return the statements with the '`' chars being in the right place to be used by impala-shell. However, we need to add the db-name in the "CREATE VIEW view_name" statement.
create_view_statement_test=`echo -e "$create_entity_statement" | grep 'CREATE VIEW'` # It needs to happen in two stages, otherwise the "grep" is not able to match multi-line statement.
if [ -n "$create_view_statement_test" ]; then
echo -e "\n'${i}' is a view, so we will save its 'create view' statement and execute it on Impala, after all tables have been created.\n"
create_view_statement=`echo -e "$create_entity_statement" | sed 's/WARN:.*//g' | sed 's/\`//g' \
| sed 's/"$/;/' | sed 's/^"//' | sed 's/\\"\\"/\"/g' | sed -e "${LOCATION_HDFS_NODE_SED_ARG}" | sed "${DATE_SED_ARG_1}" | sed "${HASH_SED_ARG_1}" | sed "${LOCATION_SED_ARG_1}" \
| sed "${DATE_SED_ARG_2}" | sed "${HASH_SED_ARG_2}" | sed "${LOCATION_SED_ARG_2}" \
| sed "${DATE_SED_ARG_3}" | sed "${HASH_SED_ARG_3}" | sed "${LOCATION_SED_ARG_3}"`
create_view_statement=`echo -e "$create_entity_statement" | sed 's/WARN:.*//g' | sed 's/"$/;/' | sed 's/^"//' | sed 's/\\"\\"/\"/g' | sed -e "${LOCATION_HDFS_NODE_SED_ARG}" | sed "s/CREATE VIEW /CREATE VIEW ${db}./"`
all_create_view_statements+=("$create_view_statement")
else
echo -e "\n'${i}' is a table, so we will check for its parquet files and create the table on Impala cluster.\n"
@ -126,12 +130,17 @@ function copydb() {
CURRENT_PRQ_FILE=`hdfs dfs -conf ${IMPALA_CONFIG_FILE} -ls -C "${IMPALA_HDFS_DB_BASE_PATH}/${db}.db/${i}/" | grep -v 'Found' | grep -v '_impala_insert_staging' | head -1`
if [ -z "$CURRENT_PRQ_FILE" ]; then # If there is not parquet-file inside.
echo -e "\nERROR: THE TABLE \"${i}\" HAD NO FILES TO GET THE SCHEMA FROM! IT'S EMPTY!\n\n"
exit 4 # Comment out when testing a DB which has such a table, just for performing this exact test-check.
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 4
fi
else
impala-shell --user ${HADOOP_USER_NAME} -i ${IMPALA_HOSTNAME} -q "create table ${db}.${i} like parquet '${CURRENT_PRQ_FILE}' stored as parquet;" |& tee error.log
log_errors=`cat error.log | grep -E "WARN|ERROR|FAILED"`
if [ -n "$log_errors" ]; then
echo -e "\n\nERROR: THERE WAS A PROBLEM WHEN CREATING TABLE '${i}'!\n\n"
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 5
fi
fi
fi
fi
@ -175,7 +184,9 @@ function copydb() {
if [[ $new_num_of_views_to_retry -eq $previous_num_of_views_to_retry ]]; then
echo -e "\n\nERROR: THE NUMBER OF VIEWS TO RETRY HAS NOT BEEN REDUCED! THE SCRIPT IS LIKELY GOING TO AN INFINITE-LOOP! EXITING..\n\n"
exit 5
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 6
fi
elif [[ $new_num_of_views_to_retry -gt 0 ]]; then
echo -e "\nTo be retried \"create_view_statements\" (${new_num_of_views_to_retry}):\n\n${all_create_view_statements[@]}\n"
else
@ -203,11 +214,14 @@ function copydb() {
else
echo -e "\n\nERROR: 1 OR MORE ENTITIES OF DB '${db}' FAILED TO BE COPIED TO IMPALA CLUSTER!\n\n"
rm -f error.log
exit 6
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 7
fi
fi
rm -f error.log
echo -e "\n\nFinished processing db: ${db}\n\n"
echo -e "\n\nFinished processing db: ${db}\n"
print_elapsed_time start_db_time
}

68
dhp-workflows/dhp-stats-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats-monitor/oozie_app/copyDataToImpalaCluster.sh
View File

@ -8,6 +8,9 @@ fi
export HADOOP_USER_NAME=$2
SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR=0
# Set the active HDFS node of OCEAN and IMPALA cluster.
OCEAN_HDFS_NODE='hdfs://nameservice1'
echo -e "\nOCEAN HDFS virtual-name which resolves automatically to the active-node: ${OCEAN_HDFS_NODE}"
@ -29,7 +32,9 @@ while [ $COUNTER -lt 3 ]; do
done
if [ -z "$IMPALA_HDFS_NODE" ]; then
echo -e "\n\nERROR: PROBLEM WHEN SETTING THE HDFS-NODE FOR IMPALA CLUSTER! | AFTER ${COUNTER} RETRIES.\n\n"
exit 1
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 1
fi
fi
echo -e "Active IMPALA HDFS Node: ${IMPALA_HDFS_NODE} , after ${COUNTER} retries.\n\n"
@ -38,26 +43,25 @@ IMPALA_CONFIG_FILE='/etc/impala_cluster/hdfs-site.xml'
IMPALA_HDFS_DB_BASE_PATH="${IMPALA_HDFS_NODE}/user/hive/warehouse"
# Set sed arguments.
LOCATION_HDFS_NODE_SED_ARG="s|${OCEAN_HDFS_NODE}|${IMPALA_HDFS_NODE}|g" # This requires to be used with "sed -e" in order to have the "|" delimiter (as the "/" conflicts with the URIs)
# Set the SED command arguments for column-names with reserved words:
DATE_SED_ARG_1='s/[[:space:]]\date[[:space:]]/\`date\`/g'
DATE_SED_ARG_2='s/\.date,/\.\`date\`,/g' # the "date" may be part of a larger field name like "datestamp" or "date_aggregated", so we need to be careful with what we are replacing.
DATE_SED_ARG_3='s/\.date[[:space:]]/\.\`date\` /g'
HASH_SED_ARG_1='s/[[:space:]]\hash[[:space:]]/\`hash\`/g'
HASH_SED_ARG_2='s/\.hash,/\.\`hash\`,/g'
HASH_SED_ARG_3='s/\.hash[[:space:]]/\.\`hash\` /g'
LOCATION_SED_ARG_1='s/[[:space:]]\location[[:space:]]/\`location\`/g'
LOCATION_SED_ARG_2='s/\.location,/\.\`location\`,/g'
LOCATION_SED_ARG_3='s/\.location[[:space:]]/\.\`location\` /g'
function print_elapsed_time()
{
start_time=$1
end_time=$(date +%s)
elapsed_time=$(($end_time-$start_time))
hours=$((elapsed_time / 3600))
minutes=$(((elapsed_time % 3600) / 60))
seconds=$((elapsed_time % 60))
printf "\nElapsed time: %02d:%02d:%02d\n\n" $hours $minutes $seconds
}
function copydb() {
db=$1
start_db_time=$(date +%s)
echo -e "\nStart processing db: '${db}'..\n"
# Delete the old DB from Impala cluster (if exists).
@ -66,7 +70,9 @@ function copydb() {
if [ -n "$log_errors" ]; then
echo -e "\n\nERROR: THERE WAS A PROBLEM WHEN DROPPING THE OLD DATABASE! EXITING...\n\n"
rm -f error.log
exit 2
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 2
fi
fi
echo -e "\n\nCopying files of '${db}', from Ocean to Impala cluster..\n"
@ -90,7 +96,9 @@ function copydb() {
else
echo -e "\n\nERROR: FAILED TO TRANSFER THE FILES OF '${db}', WITH 'hadoop distcp'. GOT EXIT STATUS: $?\n\n"
rm -f error.log
exit 3
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 3
fi
fi
# In case we ever use this script for a writable DB (using inserts/updates), we should perform the following costly operation as well..
@ -108,17 +116,13 @@ function copydb() {
num_tables=0
entities_on_ocean=`hive -e "show tables in ${db};" | sed 's/WARN:.*//g'` # Get the tables and views without any potential the "WARN" logs.
for i in ${entities_on_ocean[@]}; do # Use un-quoted values, as the elemetns are single-words.
for i in ${entities_on_ocean[@]}; do # Use un-quoted values, as the elements are single-words.
# Check if this is a view by showing the create-statement where it should print "create view" for a view, not the "create table". Unfortunately, there is no "show views" command.
create_entity_statement=`hive -e "show create table ${db}.${i};"` # It needs to happen in two stages, otherwise the "grep" is not able to match multi-line statement.
create_view_statement_test=`echo -e "$create_entity_statement" | grep 'CREATE VIEW'`
create_entity_statement=`hive --database ${db} -e "show create table ${i};"` # We need to use the "--database", instead of including it inside the query, in order to return the statements with the '`' chars being in the right place to be used by impala-shell. However, we need to add the db-name in the "CREATE VIEW view_name" statement.
create_view_statement_test=`echo -e "$create_entity_statement" | grep 'CREATE VIEW'` # It needs to happen in two stages, otherwise the "grep" is not able to match multi-line statement.
if [ -n "$create_view_statement_test" ]; then
echo -e "\n'${i}' is a view, so we will save its 'create view' statement and execute it on Impala, after all tables have been created.\n"
create_view_statement=`echo -e "$create_entity_statement" | sed 's/WARN:.*//g' | sed 's/\`//g' \
| sed 's/"$/;/' | sed 's/^"//' | sed 's/\\"\\"/\"/g' | sed -e "${LOCATION_HDFS_NODE_SED_ARG}" | sed "${DATE_SED_ARG_1}" | sed "${HASH_SED_ARG_1}" | sed "${LOCATION_SED_ARG_1}" \
| sed "${DATE_SED_ARG_2}" | sed "${HASH_SED_ARG_2}" | sed "${LOCATION_SED_ARG_2}" \
| sed "${DATE_SED_ARG_3}" | sed "${HASH_SED_ARG_3}" | sed "${LOCATION_SED_ARG_3}"`
create_view_statement=`echo -e "$create_entity_statement" | sed 's/WARN:.*//g' | sed 's/"$/;/' | sed 's/^"//' | sed 's/\\"\\"/\"/g' | sed -e "${LOCATION_HDFS_NODE_SED_ARG}" | sed "s/CREATE VIEW /CREATE VIEW ${db}./"`
all_create_view_statements+=("$create_view_statement")
else
echo -e "\n'${i}' is a table, so we will check for its parquet files and create the table on Impala cluster.\n"
@ -126,12 +130,17 @@ function copydb() {
CURRENT_PRQ_FILE=`hdfs dfs -conf ${IMPALA_CONFIG_FILE} -ls -C "${IMPALA_HDFS_DB_BASE_PATH}/${db}.db/${i}/" | grep -v 'Found' | grep -v '_impala_insert_staging' | head -1`
if [ -z "$CURRENT_PRQ_FILE" ]; then # If there is not parquet-file inside.
echo -e "\nERROR: THE TABLE \"${i}\" HAD NO FILES TO GET THE SCHEMA FROM! IT'S EMPTY!\n\n"
exit 4 # Comment out when testing a DB which has such a table, just for performing this exact test-check.
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 4
fi
else
impala-shell --user ${HADOOP_USER_NAME} -i ${IMPALA_HOSTNAME} -q "create table ${db}.${i} like parquet '${CURRENT_PRQ_FILE}' stored as parquet;" |& tee error.log
log_errors=`cat error.log | grep -E "WARN|ERROR|FAILED"`
if [ -n "$log_errors" ]; then
echo -e "\n\nERROR: THERE WAS A PROBLEM WHEN CREATING TABLE '${i}'!\n\n"
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 5
fi
fi
fi
fi
@ -175,7 +184,9 @@ function copydb() {
if [[ $new_num_of_views_to_retry -eq $previous_num_of_views_to_retry ]]; then
echo -e "\n\nERROR: THE NUMBER OF VIEWS TO RETRY HAS NOT BEEN REDUCED! THE SCRIPT IS LIKELY GOING TO AN INFINITE-LOOP! EXITING..\n\n"
exit 5
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 6
fi
elif [[ $new_num_of_views_to_retry -gt 0 ]]; then
echo -e "\nTo be retried \"create_view_statements\" (${new_num_of_views_to_retry}):\n\n${all_create_view_statements[@]}\n"
else
@ -203,11 +214,14 @@ function copydb() {
else
echo -e "\n\nERROR: 1 OR MORE ENTITIES OF DB '${db}' FAILED TO BE COPIED TO IMPALA CLUSTER!\n\n"
rm -f error.log
exit 6
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 7
fi
fi
rm -f error.log
echo -e "\n\nFinished processing db: ${db}\n\n"
echo -e "\n\nFinished processing db: ${db}\n"
print_elapsed_time start_db_time
}

72
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/copyDataToImpalaCluster.sh
View File

@ -6,6 +6,8 @@ then
ln -sfn ${PYTHON_EGG_CACHE}${link_folder} ${link_folder}
fi
SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR=0
# Set the active HDFS node of OCEAN and IMPALA cluster.
OCEAN_HDFS_NODE='hdfs://nameservice1'
@ -28,7 +30,9 @@ while [ $COUNTER -lt 3 ]; do
done
if [ -z "$IMPALA_HDFS_NODE" ]; then
echo -e "\n\nERROR: PROBLEM WHEN SETTING THE HDFS-NODE FOR IMPALA CLUSTER! | AFTER ${COUNTER} RETRIES.\n\n"
exit 1
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 1
fi
fi
echo -e "Active IMPALA HDFS Node: ${IMPALA_HDFS_NODE} , after ${COUNTER} retries.\n\n"
@ -40,26 +44,26 @@ IMPALA_HDFS_DB_BASE_PATH="${IMPALA_HDFS_NODE}/user/hive/warehouse"
# Set sed arguments.
LOCATION_HDFS_NODE_SED_ARG="s|${OCEAN_HDFS_NODE}|${IMPALA_HDFS_NODE}|g" # This requires to be used with "sed -e" in order to have the "|" delimiter (as the "/" conflicts with the URIs)
# Set the SED command arguments for column-names with reserved words:
DATE_SED_ARG_1='s/[[:space:]]\date[[:space:]]/\`date\`/g'
DATE_SED_ARG_2='s/\.date,/\.\`date\`,/g' # the "date" may be part of a larger field name like "datestamp" or "date_aggregated", so we need to be careful with what we are replacing.
DATE_SED_ARG_3='s/\.date[[:space:]]/\.\`date\` /g'
HASH_SED_ARG_1='s/[[:space:]]\hash[[:space:]]/\`hash\`/g'
HASH_SED_ARG_2='s/\.hash,/\.\`hash\`,/g'
HASH_SED_ARG_3='s/\.hash[[:space:]]/\.\`hash\` /g'
LOCATION_SED_ARG_1='s/[[:space:]]\location[[:space:]]/\`location\`/g'
LOCATION_SED_ARG_2='s/\.location,/\.\`location\`,/g'
LOCATION_SED_ARG_3='s/\.location[[:space:]]/\.\`location\` /g'
export HADOOP_USER_NAME=$6
export PROD_USAGE_STATS_DB="openaire_prod_usage_stats"
function print_elapsed_time()
{
start_time=$1
end_time=$(date +%s)
elapsed_time=$(($end_time-$start_time))
hours=$((elapsed_time / 3600))
minutes=$(((elapsed_time % 3600) / 60))
seconds=$((elapsed_time % 60))
printf "\nElapsed time: %02d:%02d:%02d\n\n" $hours $minutes $seconds
}
function copydb() {
db=$1
start_db_time=$(date +%s)
echo -e "\nStart processing db: '${db}'..\n"
# Delete the old DB from Impala cluster (if exists).
@ -68,7 +72,9 @@ function copydb() {
if [ -n "$log_errors" ]; then
echo -e "\n\nERROR: THERE WAS A PROBLEM WHEN DROPPING THE OLD DATABASE! EXITING...\n\n"
rm -f error.log
exit 2
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 2
fi
fi
echo -e "\n\nCopying files of '${db}', from Ocean to Impala cluster..\n"
@ -92,7 +98,9 @@ function copydb() {
else
echo -e "\n\nERROR: FAILED TO TRANSFER THE FILES OF '${db}', WITH 'hadoop distcp'. GOT EXIT STATUS: $?\n\n"
rm -f error.log
exit 3
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 3
fi
fi
# In case we ever use this script for a writable DB (using inserts/updates), we should perform the following costly operation as well..
@ -110,17 +118,13 @@ function copydb() {
num_tables=0
entities_on_ocean=`hive -e "show tables in ${db};" | sed 's/WARN:.*//g'` # Get the tables and views without any potential the "WARN" logs.
for i in ${entities_on_ocean[@]}; do # Use un-quoted values, as the elemetns are single-words.
for i in ${entities_on_ocean[@]}; do # Use un-quoted values, as the elements are single-words.
# Check if this is a view by showing the create-statement where it should print "create view" for a view, not the "create table". Unfortunately, there is no "show views" command.
create_entity_statement=`hive -e "show create table ${db}.${i};"` # It needs to happen in two stages, otherwise the "grep" is not able to match multi-line statement.
create_view_statement_test=`echo -e "$create_entity_statement" | grep 'CREATE VIEW'`
create_entity_statement=`hive --database ${db} -e "show create table ${i};"` # We need to use the "--database", instead of including it inside the query, in order to return the statements with the '`' chars being in the right place to be used by impala-shell. However, we need to add the db-name in the "CREATE VIEW view_name" statement.
create_view_statement_test=`echo -e "$create_entity_statement" | grep 'CREATE VIEW'` # It needs to happen in two stages, otherwise the "grep" is not able to match multi-line statement.
if [ -n "$create_view_statement_test" ]; then
echo -e "\n'${i}' is a view, so we will save its 'create view' statement and execute it on Impala, after all tables have been created.\n"
create_view_statement=`echo -e "$create_entity_statement" | sed 's/WARN:.*//g' | sed 's/\`//g' \
| sed 's/"$/;/' | sed 's/^"//' | sed 's/\\"\\"/\"/g' | sed -e "${LOCATION_HDFS_NODE_SED_ARG}" | sed "${DATE_SED_ARG_1}" | sed "${HASH_SED_ARG_1}" | sed "${LOCATION_SED_ARG_1}" \
| sed "${DATE_SED_ARG_2}" | sed "${HASH_SED_ARG_2}" | sed "${LOCATION_SED_ARG_2}" \
| sed "${DATE_SED_ARG_3}" | sed "${HASH_SED_ARG_3}" | sed "${LOCATION_SED_ARG_3}"`
create_view_statement=`echo -e "$create_entity_statement" | sed 's/WARN:.*//g' | sed 's/"$/;/' | sed 's/^"//' | sed 's/\\"\\"/\"/g' | sed -e "${LOCATION_HDFS_NODE_SED_ARG}" | sed "s/CREATE VIEW /CREATE VIEW ${db}./"`
all_create_view_statements+=("$create_view_statement")
else
echo -e "\n'${i}' is a table, so we will check for its parquet files and create the table on Impala cluster.\n"
@ -128,12 +132,17 @@ function copydb() {
CURRENT_PRQ_FILE=`hdfs dfs -conf ${IMPALA_CONFIG_FILE} -ls -C "${IMPALA_HDFS_DB_BASE_PATH}/${db}.db/${i}/" | grep -v 'Found' | grep -v '_impala_insert_staging' | head -1`
if [ -z "$CURRENT_PRQ_FILE" ]; then # If there is not parquet-file inside.
echo -e "\nERROR: THE TABLE \"${i}\" HAD NO FILES TO GET THE SCHEMA FROM! IT'S EMPTY!\n\n"
exit 4 # Comment out when testing a DB which has such a table, just for performing this exact test-check.
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 4
fi
else
impala-shell --user ${HADOOP_USER_NAME} -i ${IMPALA_HOSTNAME} -q "create table ${db}.${i} like parquet '${CURRENT_PRQ_FILE}' stored as parquet;" |& tee error.log
log_errors=`cat error.log | grep -E "WARN|ERROR|FAILED"`
if [ -n "$log_errors" ]; then
echo -e "\n\nERROR: THERE WAS A PROBLEM WHEN CREATING TABLE '${i}'!\n\n"
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 5
fi
fi
fi
fi
@ -177,7 +186,9 @@ function copydb() {
if [[ $new_num_of_views_to_retry -eq $previous_num_of_views_to_retry ]]; then
echo -e "\n\nERROR: THE NUMBER OF VIEWS TO RETRY HAS NOT BEEN REDUCED! THE SCRIPT IS LIKELY GOING TO AN INFINITE-LOOP! EXITING..\n\n"
exit 5
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 6
fi
elif [[ $new_num_of_views_to_retry -gt 0 ]]; then
echo -e "\nTo be retried \"create_view_statements\" (${new_num_of_views_to_retry}):\n\n${all_create_view_statements[@]}\n"
else
@ -205,11 +216,14 @@ function copydb() {
else
echo -e "\n\nERROR: 1 OR MORE ENTITIES OF DB '${db}' FAILED TO BE COPIED TO IMPALA CLUSTER!\n\n"
rm -f error.log
exit 6
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 7
fi
fi
rm -f error.log
echo -e "\n\nFinished processing db: ${db}\n\n"
echo -e "\n\nFinished processing db: ${db}\n"
print_elapsed_time start_db_time
}
STATS_DB=$1
@ -233,6 +247,6 @@ copydb $MONITOR_DB'_ris_tail'
contexts="knowmad::other dh-ch::other enermaps::other gotriple::other neanias-atmospheric::other rural-digital-europe::other covid-19::other aurora::other neanias-space::other north-america-studies::other north-american-studies::other eutopia::other"
for i in ${contexts}
do
tmp=`echo "$i" | sed 's/'-'/'_'/g' | sed 's/'::'/'_'/g'`
tmp=`echo "$i" | sed 's/'-'/'_'/g' | sed 's/'::'/'_'/g'`
copydb ${MONITOR_DB}'_'${tmp}
done

1
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step14.sql
View File

@ -1,3 +1,4 @@
set mapred.job.queue.name=analytics;
------------------------------------------------------
------------------------------------------------------
-- Additional relations

4
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step15.sql
View File

@ -1,3 +1,5 @@
set mapred.job.queue.name=analytics;
------------------------------------------------------
------------------------------------------------------
-- Additional relations
@ -104,4 +106,4 @@ rel.properties[1].value apc_currency
from ${openaire_db_name}.relation rel
join ${openaire_db_name}.organization o on o.id=rel.source
join ${openaire_db_name}.result r on r.id=rel.target
where rel.subreltype = 'affiliation' and rel.datainfo.deletedbyinference = false and size(rel.properties)>0; /*EOS*/
where rel.subreltype = 'affiliation' and rel.datainfo.deletedbyinference = false and size(rel.properties)>0; /*EOS*/

2
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step15_5.sql
View File

@ -1,3 +1,5 @@
set mapred.job.queue.name=analytics;
-------------------------------------------
--- Extra tables, mostly used by indicators

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

@ -249,7 +249,7 @@ create table if not exists ${stats_db_name}.indi_pub_gold_oa stored as parquet a
left semi join dd on dd.id=pd.datasource
union all
select ra.id, 1 as is_gold
from ${stats_db_name}.result_accessroute ra on ra.id = pd.id where ra.accessroute = 'gold') tmp on tmp.id=pd.id; /*EOS*/
from ${stats_db_name}.result_accessroute ra where ra.accessroute = 'gold') tmp on tmp.id=pd.id; /*EOS*/
drop table if exists ${stats_db_name}.indi_pub_hybrid_oa_with_cc purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_pub_hybrid_oa_with_cc stored as parquet as
@ -294,7 +294,7 @@ left outer join (
join ${stats_db_name}.indi_pub_gold_oa indi_gold on indi_gold.id=p.id
left outer join ${stats_db_name}.result_accessroute ra on ra.id=p.id
where indi_gold.is_gold=0 and
((d.type like '%Journal%' and ri.accessright not in ('Closed Access', 'Restricted', 'Not Available') and ri.license is not null) or ra.accessroute='hybrid')) tmp on pd.i=tmp.id; /*EOS*/
((d.type like '%Journal%' and ri.accessright not in ('Closed Access', 'Restricted', 'Not Available') and ri.license is not null) or ra.accessroute='hybrid')) tmp on p.id=tmp.id; /*EOS*/
drop table if exists ${stats_db_name}.indi_org_fairness purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_org_fairness stored as parquet as
@ -1008,14 +1008,14 @@ left outer join (
drop table if exists ${stats_db_name}.result_country purge; /*EOS*/
create table ${stats_db_name}.result_country stored as parquet as
select distinct *
select distinct id, country
from (
select ro.id, o.country
from ${stats_db_name}.result_organization ro
left outer join ${stats_db_name}.organization o on o.id=ro.organization
union all
select rp.id, f.country
from ${stats_db_name}.result_projects
from ${stats_db_name}.result_projects rp
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
) rc

2
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16_1-definitions.sql
View File

@ -1,3 +1,5 @@
set mapred.job.queue.name=analytics;
----------------------------------------------------
-- Shortcuts for various definitions in stats db ---
----------------------------------------------------

2
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16_5.sql
View File

@ -1,3 +1,5 @@
set mapred.job.queue.name=analytics;
-- replace the creation of the result view to include the boolean fields from the previous tables (green, gold,
-- peer reviewed)
drop table if exists ${stats_db_name}.result_tmp; /*EOS*/

4
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step2.sql
View File

@ -1,3 +1,5 @@
set mapred.job.queue.name=analytics;
--------------------------------------------------------------
--------------------------------------------------------------
-- Publication table/view and Publication related tables/views
@ -111,4 +113,4 @@ SELECT /*+ COALESCE(100) */ substr(p.id, 4) AS id, xpath_string(citation.value,
FROM ${openaire_db_name}.publication p
lateral view explode(p.extrainfo) citations AS citation
WHERE xpath_string(citation.value, "//citation/id[@type='openaire']/@value") != ""
and p.datainfo.deletedbyinference = false and p.datainfo.invisible=false; /*EOS*/
and p.datainfo.deletedbyinference = false and p.datainfo.invisible=false;

9
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step7.sql
View File

@ -129,11 +129,14 @@ create table ${stats_db_name}.result_fos stored as parquet as
with
lvl1 as (select id, topic from ${stats_db_name}.result_topics where topic like '__ %' and type='Fields of Science and Technology classification'),
lvl2 as (select id, topic from ${stats_db_name}.result_topics where topic like '____ %' and type='Fields of Science and Technology classification'),
lvl3 as (select id, topic from ${stats_db_name}.result_topics where topic like '______ %' and type='Fields of Science and Technology classification')
select /*+ COALESCE(100) */ lvl1.id, lvl1.topic as lvl1, lvl2.topic as lvl2, lvl3.topic as lvl3
lvl3 as (select id, topic from ${stats_db_name}.result_topics where topic like '______ %' and type='Fields of Science and Technology classification'),
lvl4 as (select id, topic from ${stats_db_name}.result_topics where topic like '________ %' and type='Fields of Science and Technology classification')
select /*+ COALESCE(100) */ lvl1.id, lvl1.topic as lvl1, lvl2.topic as lvl2, lvl3.topic as lvl3, lvl4.topic as lvl4
from lvl1
join lvl2 on lvl1.id=lvl2.id and substr(lvl2.topic, 1, 2)=substr(lvl1.topic, 1, 2)
join lvl3 on lvl3.id=lvl1.id and substr(lvl3.topic, 1, 4)=substr(lvl2.topic, 1, 4); /*EOS*/
join lvl3 on lvl3.id=lvl1.id and substr(lvl3.topic, 1, 4)=substr(lvl2.topic, 1, 4)
join lvl4 on lvl4.id=lvl1.id and substr(lvl4.topic, 1, 6)=substr(lvl3.topic, 1, 6);
DROP TABLE IF EXISTS ${stats_db_name}.result_organization purge; /*EOS*/

3
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/workflow.xml
View File

@ -568,6 +568,7 @@
${sparkClusterOpts}
${sparkResourceOpts}
${sparkApplicationOpts}
--queue analytics
</spark-opts>
<arg>--hiveMetastoreUris</arg><arg>${hive_metastore_uris}</arg>
<arg>--sql</arg><arg>eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16-createIndicatorsTables.sql</arg>
@ -784,4 +785,4 @@
</action>
<end name="End"/>
</workflow-app>
</workflow-app>