What’s the difference between sum_irate and sum(rate())?
Difference between sum_irate and sum(rate())? I uses two metrics node_namespace_pod_container:container_cpu_usage_seconds_total:sum_irate and container_cpu_usage_seconds_total with the sum, rate function sum_irate node_namespace_pod_container:container_cpu_usage_seconds_total:sum_irate{namespace=”$namespace”, pod=~”$PodPrefix-.*”, container!=””} sum(rate()) sum(rate(container_cpu_usage_seconds_total{namespace=”$namespace”, pod=~”$PodPrefix-.*”, container!=””}[1m])) by (pod) These show different values. Also, both shows exceed 100% CPU Util. I wonder what is difference between them. I expected two metric result in same value. But it’s […]
How can exceed CPU utilization 100% of Pod even though set request, limit 1000m?
Qeustion1. Why does cpu util exceed 100%? deployment.yaml I restrict pod CPU utilization as 1000m. resources: requests: cpu: 1000m limits: cpu: 1000m promql node_namespace_pod_container:container_cpu_usage_seconds_total:sum_irate{namespace=”$namespace”, pod=~”$PodPrefix-.*”, container!=””} Grafana I added dashboard naming Pod CPU Usage. It shows each pod CPU Utilization. It seems that some Pod CPU util exceeds limit (100%). 140%, even 215% . How […]
What is the difference between namespace centred metrics and metrics aggregated based on name space in Prometheus Query
I am trying to get namespace cpu usages usign the following query
Grafana + Prometheus don’t show data from the previous day
Every morning I see the same problem. Graphs of CPU, Disk and Memory of Kubernetes nodes start from 6 AM/8:30 AM but I can’t see previous data.