How would I graph cpu usage in grafana using Prometheus and the collectd exporter?

grafana prometheus collectd promql
Drew Pierce picture Drew Pierce · Aug 1, 2018 · Viewed 11.4k times · Source

New to promql and I'm trying to graph cpu usage over time and I've found the following example getting a static percent of usage but how would I convert it to a rate for grafana?

100 * (1 - sum(collectd_cpu_total{type="idle"}) by (exported_instance) / sum(collectd_cpu_total) by (exported_instance))

I got this example from https://github.com/Naugrimm/promql-examples

Answer

Hang picture Hang · Aug 2, 2018

Since you are using Prometheus already, why don't you use Prometheus' node_exporter? https://www.robustperception.io/understanding-machine-cpu-usage

100 - (avg by (instance) (irate(node_cpu_seconds_total{job="node",mode="idle"}[5m])) * 100)

Related questions

How do I write a Prometheus query that returns the value of a label?

I'm making a Grafana dashboard and want a panel that reports the latest version of our app. The version is reported as a label in the app_version (say) metric like so: app_version_updated{instance="eu99",version="1.5.0-abcdefg"} I've …

Read More
How can I make a Grafana template with a variable reference another variable using Prometheus as a datasource?

I have a Grafana dashboard with template variables for services and instances. When I select a service how can I make it filter the second template variable list based on the first?

Read More
Prometheus query to count unique label values

I want to count number of unique label values. Kind of like select count (distinct a) from hello_info For example if my metric 'hello_info' has labels a and b. I want to count number of unique a's. Here …

Read More