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

Merge pull request 'changes in copy script - beta2master' (#439) from antonis.lempesis/dnet-hadoop:beta into beta_to_master_may2024 

Reviewed-on: #439
This commit is contained in:
Claudio Atzori 2024-05-29 14:29:26 +02:00
parent 14f275ffaf e3f28338c1
commit b703f94f09
5 changed files with 243 additions and 133 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

90
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: '${db}'..\n"
# Delete the old DB from Impala cluster (if exists).
@ -67,7 +70,11 @@ function copydb() {
if [ -n "$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
else
return 2
fi
fi
echo -e "\n\nCopying files of '${db}', from Ocean to Impala cluster..\n"
@ -85,17 +92,30 @@ function copydb() {
-pb \
${OCEAN_HDFS_NODE}/user/hive/warehouse/${db}.db ${IMPALA_HDFS_DB_BASE_PATH}
# Check the exit status of the "hadoop distcp" command.
if [ $? -eq 0 ]; then
echo -e "\nSuccessfully copied the files of '${db}'.\n"
if [ $? -eq 0 ]; then # Check the exit status of the "hadoop distcp" command.
echo -e "\nSuccessfully copied the files of '${db}' from Ocean to Impala cluster.\n"
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
else
return 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..
#hdfs dfs -conf ${IMPALA_CONFIG_FILE} -chmod -R 777 ${TEMP_SUBDIR_FULLPATH}/${db}.db
# Give WRITE and EXECUTE permissions to the DBs' directory only, in order to be able to create more tables later, on top of that DB.
hdfs dfs -conf ${IMPALA_CONFIG_FILE} -chmod u+wx ${IMPALA_HDFS_DB_BASE_PATH}/${db}.db
# In case we ever use this script for a writable DB (using inserts/updates), we should perform the costly recursive operation as well, using the "-R" param.
if [ $? -ne 0 ]; then # Check the exit status..
echo -e "\n\nERROR: FAILED TO ASSIGN WRITE AND EXECUTE PERMISSIONS TO THE DIRECTORY OF DB: '${db}'. GOT EXIT STATUS: $?\n\n"
rm -f error.log
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 4
else
return 4
fi
fi
echo -e "\nCreating schema for db: '${db}'\n"
@ -109,17 +129,13 @@ 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 +143,17 @@ 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 5
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 6
fi
fi
fi
fi
@ -176,7 +197,9 @@ 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 7
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${all_create_view_statements[@]}\n"
else
@ -204,11 +227,14 @@ function copydb() {
else
echo -e "\n\nERROR: 1 OR MORE ENTITIES OF DB '${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 8
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
}

91
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: '${db}'..\n"
# Delete the old DB from Impala cluster (if exists).
@ -66,7 +70,11 @@ function copydb() {
if [ -n "$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
else
return 2
fi
fi
echo -e "\n\nCopying files of '${db}', from Ocean to Impala cluster..\n"
@ -84,17 +92,30 @@ function copydb() {
-pb \
${OCEAN_HDFS_NODE}/user/hive/warehouse/${db}.db ${IMPALA_HDFS_DB_BASE_PATH}
# Check the exit status of the "hadoop distcp" command.
if [ $? -eq 0 ]; then
echo -e "\nSuccessfully copied the files of '${db}'.\n"
if [ $? -eq 0 ]; then # Check the exit status of the "hadoop distcp" command.
echo -e "\nSuccessfully copied the files of '${db}' from Ocean to Impala cluster.\n"
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
else
return 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..
#hdfs dfs -conf ${IMPALA_CONFIG_FILE} -chmod -R 777 ${TEMP_SUBDIR_FULLPATH}/${db}.db
# Give WRITE and EXECUTE permissions to the DBs' directory only, in order to be able to create more tables later, on top of that DB.
hdfs dfs -conf ${IMPALA_CONFIG_FILE} -chmod u+wx ${IMPALA_HDFS_DB_BASE_PATH}/${db}.db
# In case we ever use this script for a writable DB (using inserts/updates), we should perform the costly recursive operation as well, using the "-R" param.
if [ $? -ne 0 ]; then # Check the exit status..
echo -e "\n\nERROR: FAILED TO ASSIGN WRITE AND EXECUTE PERMISSIONS TO THE DIRECTORY OF DB: '${db}'. GOT EXIT STATUS: $?\n\n"
rm -f error.log
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 4
else
return 4
fi
fi
echo -e "\nCreating schema for db: '${db}'\n"
@ -108,17 +129,13 @@ 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 +143,17 @@ 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 5
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 6
fi
fi
fi
fi
@ -175,7 +197,9 @@ 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 7
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${all_create_view_statements[@]}\n"
else
@ -203,11 +227,14 @@ function copydb() {
else
echo -e "\n\nERROR: 1 OR MORE ENTITIES OF DB '${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 8
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
}

91
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: '${db}'..\n"
# Delete the old DB from Impala cluster (if exists).
@ -66,7 +70,11 @@ function copydb() {
if [ -n "$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
else
return 2
fi
fi
echo -e "\n\nCopying files of '${db}', from Ocean to Impala cluster..\n"
@ -84,17 +92,30 @@ function copydb() {
-pb \
${OCEAN_HDFS_NODE}/user/hive/warehouse/${db}.db ${IMPALA_HDFS_DB_BASE_PATH}
# Check the exit status of the "hadoop distcp" command.
if [ $? -eq 0 ]; then
echo -e "\nSuccessfully copied the files of '${db}'.\n"
if [ $? -eq 0 ]; then # Check the exit status of the "hadoop distcp" command.
echo -e "\nSuccessfully copied the files of '${db}' from Ocean to Impala cluster.\n"
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
else
return 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..
#hdfs dfs -conf ${IMPALA_CONFIG_FILE} -chmod -R 777 ${TEMP_SUBDIR_FULLPATH}/${db}.db
# Give WRITE and EXECUTE permissions to the DBs' directory only, in order to be able to create more tables later, on top of that DB.
hdfs dfs -conf ${IMPALA_CONFIG_FILE} -chmod u+wx ${IMPALA_HDFS_DB_BASE_PATH}/${db}.db
# In case we ever use this script for a writable DB (using inserts/updates), we should perform the costly recursive operation as well, using the "-R" param.
if [ $? -ne 0 ]; then # Check the exit status..
echo -e "\n\nERROR: FAILED TO ASSIGN WRITE AND EXECUTE PERMISSIONS TO THE DIRECTORY OF DB: '${db}'. GOT EXIT STATUS: $?\n\n"
rm -f error.log
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 4
else
return 4
fi
fi
echo -e "\nCreating schema for db: '${db}'\n"
@ -108,17 +129,13 @@ 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 +143,17 @@ 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 5
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 6
fi
fi
fi
fi
@ -175,7 +197,9 @@ 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 7
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${all_create_view_statements[@]}\n"
else
@ -203,11 +227,14 @@ function copydb() {
else
echo -e "\n\nERROR: 1 OR MORE ENTITIES OF DB '${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 8
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
}

95
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: '${db}'..\n"
# Delete the old DB from Impala cluster (if exists).
@ -68,7 +72,11 @@ function copydb() {
if [ -n "$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
else
return 2
fi
fi
echo -e "\n\nCopying files of '${db}', from Ocean to Impala cluster..\n"
@ -86,17 +94,30 @@ function copydb() {
-pb \
${OCEAN_HDFS_NODE}/user/hive/warehouse/${db}.db ${IMPALA_HDFS_DB_BASE_PATH}
# Check the exit status of the "hadoop distcp" command.
if [ $? -eq 0 ]; then
echo -e "\nSuccessfully copied the files of '${db}'.\n"
if [ $? -eq 0 ]; then # Check the exit status of the "hadoop distcp" command.
echo -e "\nSuccessfully copied the files of '${db}' from Ocean to Impala cluster.\n"
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
else
return 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..
#hdfs dfs -conf ${IMPALA_CONFIG_FILE} -chmod -R 777 ${TEMP_SUBDIR_FULLPATH}/${db}.db
# Give WRITE and EXECUTE permissions to the DBs' directory only, in order to be able to create more tables later, on top of that DB.
hdfs dfs -conf ${IMPALA_CONFIG_FILE} -chmod u+wx ${IMPALA_HDFS_DB_BASE_PATH}/${db}.db
# In case we ever use this script for a writable DB (using inserts/updates), we should perform the costly recursive operation as well, using the "-R" param.
if [ $? -ne 0 ]; then # Check the exit status..
echo -e "\n\nERROR: FAILED TO ASSIGN WRITE AND EXECUTE PERMISSIONS TO THE DIRECTORY OF DB: '${db}'. GOT EXIT STATUS: $?\n\n"
rm -f error.log
if [[ SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR -eq 1 ]]; then
exit 4
else
return 4
fi
fi
echo -e "\nCreating schema for db: '${db}'\n"
@ -110,17 +131,13 @@ 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 +145,17 @@ 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 5
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 6
fi
fi
fi
fi
@ -177,7 +199,9 @@ 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 7
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${all_create_view_statements[@]}\n"
else
@ -205,11 +229,14 @@ function copydb() {
else
echo -e "\n\nERROR: 1 OR MORE ENTITIES OF DB '${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 8
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 +260,6 @@ 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

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 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 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);
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;