Monitor Atomic host services through ansible

Question

Is there any way to continuous monitor some host level services docker, kubelet, kube-proxy, flanneld etc. using ansible or prometheus?

I don't want to install anything in my Atomic.

Answer 1

Ansible is a configuration automation and management tool. It is not intended for monitoring. So, most likely no.

Answer 2

This is where the monitoring stack gets complex. Although Prometheus is good at what it does it doesn't do everything you might need an agent for. This includes custom time series metrics, logs, or deeper diagnostics or application performance monitoring. Often times you need to combine these tools, which means you may want to run an exporter like statsd, telegraf, Dropwizard, and others. The issue is that each one of these creates more footprint and complexity, not just in your containers, but across what you have to manage. This is the reason some folks choose to use a SaaS service for monitoring. There are loads of them out there, each with pros and cons.

Answer 3

prometheus is an awesome monitoring tool, but it pulls its metrics from the hosts you're monitoring. If the services you're trying to monitor expose these metrics, then you dont need to install anything. But most likely they dont, so you'll need to install some kind of metrics exporter to get host-level metrics. We use telegraf, but there are other alternatives.