PNDA-METRICS

• PNDA cluster writes metrics to graphite. In release 4.0, Spark metrics from all executors will sink to graphite.
• Influx DB is installed and configured to hold all metrics required by the back end systems.
• A telegraf agent is used to collect data from JMX sources (Kafka brokers) through Jolokia plugin.
• Telegraf agent is configured to output the collected data to Influx DB.
• Influx DB is also configured to source and store metrics from graphite.
• Telegraf agent can also be configured to produce data to Kafka.

Build Script

• Build server need to have access Internet, to download the packages
• User need to have existing PNDA mirror server
• User need to have access to mirror server , to copy the mirror_metrics directory created by build script to PNDA_MIRROR/mirror_metrics
• Deployment script going to use existing PNDA_MIRROR/mirror_misc packages [java package for jetty]
• Build script need to run as root

Run Build script

Step1 : clone the PNDA_METRICS

 git clone https://github.com/janselva/PNDA-METRICS.git

Step2: Run Build script

  sudo su
 
  cd build 

  ./build_metrics.sh

Step3: copy the mirror_metrics directory to pnda mirror server.

Deployment script

• Deployment server is reachable to Metrics server either directly or through bastion server.
• Metrics username provide in the metrics_env.yaml file must be a sudo user
• Metric server reachable to the mirror server , to download packages
• Metrics packages will be download only from mirror server
• Telegraf input,output and socket listeners detail provide in the metrics_env.yaml file
• Script tested in AWS environment.
• Deployment logs stored in deployment server itself

Run Deployment script

step1: update the cli/metrics_env.yaml file with proper values

Step2: run python script

    cd cli

    python metrics_cli.py

Step3: Logs will be stored in cli/logs folder