How to check incoming and outgoing metrics/writes per second for telegraf?

Telegraf
,
1

I am ingesting data from Kafka into telegraf. And my output data is going into InfluxDB. I want to find optimum values for my telegraf config:
image
This is my current configuration. How can I monitor my input and output writes ?Any suggestions on how to configure to reduce memory consumption and avoid buffer overflow?

2

There are plugins for that as well, if you want to monitor Telegraf check the Input Internal.

Is extremely easy and allows you to monitor lots of useful stuff with a bunch of query:

  • points frequency (in-out)
  • RAM consumption
  • buffered points
  • buffer limit
  • and even more, just explore a bit the data
3

Thanks for replying! One more query , these values will change with changes in config right ? Like max_undelivered messages or input interval

4

“kind of yes” is the best answer I can give you.

The input will monitor Telegraf itself, using the default gathering interval.
Some metrics will be cumulative, like the data points gathered/written (for which you want to use the non_negative_difference and derivative), a small interval just means more data points and more precision in your monitoring.

How easy is it to see the difference made in the config (like max_undelivered) is a totally different topic… it’s all up to how much data you have, at which frequency you gather them, how you visualize them, etc…

here is a screenshot of my current dashboard, so you get an idea of some of the metrics you get… there are several metrics I don’t consider

image
image1845Ă—1595 321 KB

5

Hey , could u explain me how did you create this dashboard? It looks really efficient! If you could share how the config files look, it would be much helpful.

6

From the telegraf side, it’s just the input.internal… which has almost no config.
for the dashboard itself, I might clean it up a little and add it to the Grafana community dashboards.
ill’ edit or reply again once I’ve done that

2 Likes