telegraf tailcount

Counts lines in growing file and return metrics on STDOUT.

Use with execd plugin.

Why ?

Mainly because apache ReqPerSec from mod_status is totally useless.

Usage

Invocation

Usage: telegraf-tailcount [options] file
Options:
  -i int     Interval in seconds (default 10)
  -p int     Peak interval in seconds (default 1)
  -m string  Measurement name (default 'tailcount')
  -t string  Comma-separated k=p pairs for tags (default 'file=<file>')

e.g.:

$ telegraf-tailcount -t foo=bar -i 10 -p 1 /tmp/log
tailcount,file=/tmp/log,foo=bar sum=31,max=5,min=1
tailcount,file=/tmp/log,foo=bar sum=35,max=6,min=1
...

Interval is the metric output interval on STDOUT.
Peak interval is the interval for which telegraf-tailcount will compute min, mean, median and max

So if you're looking at computing requests per second from an http log file, you'll want Peak interval to be 1.

Note that Interval must be a multiple of Peak interval, and both must be integers.

Telegraf config

This tool is intended to be used with the execd telegraf plugin.

[[inputs.execd]]
  command = ["telegraf-tailcount", "-t", "foo=bar", "-i", "10", "-p", "1", "/tmp/log"]

  # the values below are the defaults and can be omitted
  signal = "none"
  restart_delay = "10s"
  data_format = "influx"

Metrics

telegraf-tailcount outputs the following metrics:

sum: number of lines read during interval
max: maximum of lines read during a peakInterval
min: minimum of lines read during a peakInterval
median: median of lines read during a peakInterval
mean: mean of lines read during a peakInterval

devops-works / telegraf-tailcount