sabeel
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Pull Requests Activity

Small updates to the copy-operation to Impala Cluster: 

- Add a configuration-"switch" to control whether the script exits upon an error or not.
- Allow the script to exit when a table could not be created.
- Show the elapsed time for processing each database.
This commit is contained in:
Lampros Smyrnaios 2024-05-23 15:07:49 +03:00
parent c7b32bbacc
commit 68322843e2
4 changed files with 153 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

44
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 +8,7 @@ fi
export HADOOP_USER_NAME=$2
SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR=1
# Set the active HDFS node of OCEAN and IMPALA cluster.
OCEAN_HDFS_NODE='hdfs://nameservice1'
@ -30,7 +31,9 @@ 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 ${COUNTER} retries.\n\n"
@ -43,8 +46,21 @@ IMPALA_HDFS_DB_BASE_PATH="${IMPALA_HDFS_NODE}/user/hive/warehouse"
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)
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: '${db}'..\n"
# Delete the old DB from Impala cluster (if exists).
@ -53,7 +69,9 @@ function copydb() {
if [ -n "$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 '${db}', from Ocean to Impala cluster..\n"
@ -77,7 +95,9 @@ function copydb() {
else
echo -e "\n\nERROR: FAILED TO TRANSFER THE FILES OF '${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..
@ -109,12 +129,17 @@ 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
@ -158,7 +183,9 @@ 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${all_create_view_statements[@]}\n"
else
@ -186,11 +213,14 @@ function copydb() {
else
echo -e "\n\nERROR: 1 OR MORE ENTITIES OF DB '${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
}

46
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=1
# 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 ${COUNTER} retries.\n\n"
@ -42,8 +47,21 @@ IMPALA_HDFS_DB_BASE_PATH="${IMPALA_HDFS_NODE}/user/hive/warehouse"
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)
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: '${db}'..\n"
# Delete the old DB from Impala cluster (if exists).
@ -52,7 +70,9 @@ function copydb() {
if [ -n "$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 '${db}', from Ocean to Impala cluster..\n"
@ -76,7 +96,9 @@ function copydb() {
else
echo -e "\n\nERROR: FAILED TO TRANSFER THE FILES OF '${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..
@ -108,12 +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
@ -157,7 +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${all_create_view_statements[@]}\n"
else
@ -185,11 +214,14 @@ function copydb() {
else
echo -e "\n\nERROR: 1 OR MORE ENTITIES OF DB '${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
}

45
dhp-workflows/dhp-stats-monitor-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats-monitor/oozie_app/copyDataToImpalaCluster.sh
View File

@ -8,6 +8,8 @@ fi
export HADOOP_USER_NAME=$2
SHOULD_EXIT_WHOLE_SCRIPT_UPON_ERROR=1
# 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 +31,9 @@ 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 ${COUNTER} retries.\n\n"
@ -42,8 +46,21 @@ IMPALA_HDFS_DB_BASE_PATH="${IMPALA_HDFS_NODE}/user/hive/warehouse"
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)
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: '${db}'..\n"
# Delete the old DB from Impala cluster (if exists).
@ -52,7 +69,9 @@ function copydb() {
if [ -n "$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 '${db}', from Ocean to Impala cluster..\n"
@ -76,7 +95,9 @@ function copydb() {
else
echo -e "\n\nERROR: FAILED TO TRANSFER THE FILES OF '${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..
@ -108,12 +129,17 @@ 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
@ -157,7 +183,9 @@ 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${all_create_view_statements[@]}\n"
else
@ -185,11 +213,14 @@ function copydb() {
else
echo -e "\n\nERROR: 1 OR MORE ENTITIES OF DB '${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
}

47
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=1
# 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"
@ -45,8 +49,21 @@ 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: '${db}'..\n"
# Delete the old DB from Impala cluster (if exists).
@ -55,7 +72,9 @@ function copydb() {
if [ -n "$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 '${db}', from Ocean to Impala cluster..\n"
@ -79,7 +98,9 @@ function copydb() {
else
echo -e "\n\nERROR: FAILED TO TRANSFER THE FILES OF '${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..
@ -111,12 +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
@ -160,7 +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${all_create_view_statements[@]}\n"
else
@ -188,11 +216,14 @@ function copydb() {
else
echo -e "\n\nERROR: 1 OR MORE ENTITIES OF DB '${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
@ -216,6 +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