I need CPU usage as the proportion of the maximum CPU usage. # TYPE node_cpu counter node_cpu_seconds_total{cpu="0",mode="guest"} 0 node_cpu_seconds_total{cpu="0",mode="idle"} … On a Node exporters' metrics page, part of the output is: # HELP node_cpu Seconds the cpus spent in each mode. Today I want to tackle one apparently obvious thing, which is getting a graph (or numbers) of CPU utilization. High CPU load is a common cause of issues. Not sure it is Prometheus or we are trying it wrong. You can use container_spec_cpu_shares in place of container_spec_cpu_quota in the original query listed #2026 (comment) to pull what appear to be container CPU requests, but this means you can also potentially see CPU utilization over 100% if usage goes over requests.
Not sure it is Prometheus or we are trying it wrong. Both queries you cited give the current CPU usage of the namespaces in cores or CPU time (would be nice to know which), but that's not what I need. Not sure it is Prometheus or we are trying it wrong. I run a Prometheus instance that evaluates rules every 10 seconds. Please let me know if that helped.

Closed vincus opened this issue Sep 4, 2017 ... ordinary system cpu usage data.

This comment has been minimized. relative complexity on both the user and server side vs the system call for the collector. Understanding Machine CPU usage.

Understanding Machine CPU usage. If I scrape that Prometheus instance every 5s and look at irate() with a resolution that's a multiple of 10 seconds, I will only ever see either the spikes (when rules are evaluated) or the troughs, whereas the actual CPU utilization is actually the average of the two. Or the low CPU usage periods in-between spikes. A certain amount of Prometheus's query language is reasonably obvious, but once you start getting into the details and the clever tricks you wind up needing to wrap your mind around how PromQL wants you to think about its world. Now that the service is running, we have to create the Prometheus integration in order to get the metrics. Of course you can adjust the [1m] parameter (and others) as you need.
If I scrape that Prometheus instance every 5s and look at irate() with a resolution that's a multiple of 10 seconds, I will only ever see either the spikes (when rules are evaluated) or the troughs, whereas the actual CPU utilization is actually the average of the two. A certain amount of Prometheus's query language is reasonably obvious, but once you start getting into the details and the clever tricks you wind up needing to wrap your mind around how PromQL wants you to think about its world. Well, you ask prometheus for user cpu and you get it, then … Not sure it is Prometheus or we are trying it wrong. Copy link Quote reply Member ktsaou commented Sep 4, 2017. It does this by a calculation based on the idle metric of the CPU, working out the overall percentage of the other states for a CPU in a 5 minute window and presenting that data per instance.