#collectd_tool

About CollectdTool

CollectdTool.py is a simple script for aggregating the many files that the collectd CSV plugin generates into a single file. In order to find common ground between many different files, I've rolled up stats using a rolling average with a 1 minute window. The leftmost column on the final generated CSV file is elapsed_minutes because of this.

TODO

• Split into multiple files by group (riak, cpu, memory, etc)
• Integrate with R or another automatic graph generation technique
• Make the rolling average window configurable, as the Interval collectd config value could be > 1 minute, though not likely
• Make column names prettier and easier to read possibly through some regex magic

Installing Collectd

Prerequisites

yajl (Yet Another Json Library, needed for Riak Stats)

• brew install yajl
• yum install yajl
• apt-get install yajl

Install Collectd (https://collectd.org/wiki/index.php/First_steps for more information)

• brew install collectd
• yum install collectd
• apt-get install collectd

Setup / Start Collectd

git clone git://github.com/drewkerrigan/collectd_tool.git
cd collectd_tool

I've included a sample collectd.conf that can be used to track riak stats and a minimum of other interesting stats.

Linux

mv /etc/collectd/collectd.conf /etc/collectd/collectd.conf.bak
cp collectd.conf /etc/collectd/collectd.conf
/etc/init.d/collectd start

Mac OS X

sudo mkdir -p /var/lib/collectd/csv
sudo chmod 777 /var/lib/collectd/csv
collectd -C collectd.conf

Stopping Collectd

Linux

/etc/init.d/collectd stop

Mac OS X

ps -ef | grep collectd
>   501 74052 42165   0 10:30PM ttys004    0:00.05 collectd -f -C /Users/dkerrigan/src/collectd_tool/collectd.conf
kill 74052

Aggregating Statistics

python CollectdTool.py /var/lib/collectd/csv > stats_summary.csv

Cleaning Up Statistics

rm -rf  /var/lib/collectd/csv/*