Monitoring applications running in kubernetes containers

Our application writes statsd metrics. We run this application inside a pod in a deployment in kubernetes. We’d like to use telegraf to collect these statsd metrics.

What’s the right way to do this?

  1. Run telegraf in each pod, as a separate container with the application – if so, are there any examples on how to do this?
  2. Run telegraf-s
  3. Run telegraf-ds

Hi there, yup, here at influx we are using telegraf as a side car all the time.

Check out our blog here about it: https://www.influxdata.com/blog/monitoring-kubernetes-architecture/

You could configure telegraf to be a statsd collector and forward the data on.

In some recent work where I was building an k8s job, I actually decided against statsd style metrics because I did not want aggregation. I wanted each and every stat. So, I changed the telegraf side car from being a statsd collect to a socket listener that took line protocol directly. I’m not suggesting you switch, but, with a sidecar you’ll get the freedom to iterate.

1 Like