ganglia metrics

seefoods · ‎08-31-2023

Hello Everyone,

I have build this script in order to collect ganglia metrics but the size of stderr and sdtout ganglia is 0. It doesn't work.
I Have put this script on Workspace due to migration databricks all init-script should be place on Workspace not dbfs, but it doesnt work. Do you know how can i solve this issue?

#!/bin/bash

set -e # Exit on errors

# This script must run only on the driver node

if [ \$DB_IS_DRIVER != "TRUE" ]; then

exit 0

fi

if [ \$GANGLIA_METRICS_ENABLED != "true" ]; then

echo "Ganglia metrics on \$DB_CLUSTER_NAME are not enabled"

exit 0

fi

echo "Enabling collection of metrics on \$DB_CLUSTER_NAME"

cat <<'EOF' >>/tmp/gather_ganglia_metrics.sh

#!/bin/bash

ROOT_PATH="/Workspace/ganglia_metrics"

LOGS_DIR="\$ROOT_PATH/\$DB_CLUSTER_NAME"

# Assign poll interval

re='^[0-9]+$'

if [[ \$GANGLIA_METRICS_POLL_INTERVAL =~ \$re ]]; then

POLL_INTERVAL=\$GANGLIA_METRICS_POLL_INTERVAL

else

POLL_INTERVAL=15

fi

echo "Poll interval is \$POLL_INTERVAL seconds"

if [ ! -d \$LOGS_DIR ]; then

sudo mkdir -p \$LOGS_DIR

echo "directory \$LOGS_DIR is created"

fi

while true; do

LOG_TIMESTAMP=$(date '+%Y%m%d%H%M%S')

LOG_PATH="\$LOGS_DIR/\${DB_DRIVER_IP}_\${LOG_TIMESTAMP}.xml"

echo \$LOG_PATH

curl http://localhost:8652/cluster/*/{cpu_aidle,cpu_idle,cpu_nice,cpu_num,cpu_report,cpu_speed,cpu_system} >> \$LOG_PATH

sleep \$POLL_INTERVAL

done

EOF

if [ $DB_IS_DRIVER ]; then

sudo chmod a+x /tmp/gather_ganglia_metrics.sh

nohup /tmp/gather_ganglia_metrics.sh &

fi

Debayan · ‎09-03-2023

Hi, Is there any error you are getting?

Also, please tag @Debayan with your next comment which will notify me. Thanks!