search

devopsprodigy / kubegraf

Grafana-plugin for k8s' monitoring
MIT License
407 stars 45 forks source link

Update CPU usage by pod #26

Closed eug-maly closed 4 years ago

eug-maly commented 4 years ago

CPU usage by pod panel has the following query: sum(rate(container_cpu_usage_seconds_total{namespace="$namespace", pod_name=~"$pod"}[1m])) by (pod_name)

When I test the query in the prometheus ui: (sum by(pod_name) (rate(container_cpu_usage_seconds_total{pod_name="TEST_POD",image!=""}[5m])) * 100) get result: {pod_name="TEST_POD"} 43.88829601597685

the same query without sum: rate(container_cpu_usage_seconds_total{ pod_name="TEST_POD"}[5m]) 

container_cpu_usage_seconds_total{container="POD",container_name="POD",cpu="total",endpoint="https-metrics",id="/kubepods/burstable/pod937131a1-436b-11ea-9b9c-0ad04b43bf50/6865ecb90d75a7ae9316e9a131c5a9447714cff5333994c512773d4a64621fe9",image="602401143452.dkr.ecr.ap-southeast-2.amazonaws.com/eks/pause-amd64:3.1",instance="10.50.50.63:10250",job="kubelet",name="k8s_POD_TEST_POD_default_937131a1-436b-11ea-9b9c-0ad04b43bf50_0",namespace="default",node="ip-10-50-50-63.ap-southeast-2.compute.internal",pod="TEST_POD",pod_name="TEST_POD",service="prometheus-operator-kubelet"}   0.01538971

container_cpu_usage_seconds_total{container="couchbase-server",container_name="couchbase-server",cpu="total",endpoint="https-metrics",id="/kubepods/burstable/pod937131a1-436b-11ea-9b9c-0ad04b43bf50/fa9326b7237f447ab499b1d7363b537aa06c97174eb874643d5bb236a1e9c41f",image="sha256:fbaae96e8d377ee42762082d5ed9113afef0177f8b848b29b161c699d8447bfc",instance="10.50.50.63:10250",job="kubelet",name="k8s_couchbase-server_TEST_POD_default_937131a1-436b-11ea-9b9c-0ad04b43bf50_0",namespace="default",node="ip-10-50-50-63.ap-southeast-2.compute.internal",pod="TEST_POD",pod_name="TEST_POD",service="prometheus-operator-kubelet"}  255391.764905599

container_cpu_usage_seconds_total{cpu="total",endpoint="https-metrics",id="/kubepods/burstable/pod937131a1-436b-11ea-9b9c-0ad04b43bf50",instance="10.50.50.63:10250",job="kubelet",namespace="default",node="ip-10-50-50-63.ap-southeast-2.compute.internal",pod="TEST_POD",pod_name="TEST_POD",service="prometheus-operator-kubelet"}  255394.456398121

Not sure what does the last line container_cpu_usage_seconds_total{cpu="total"... mean, but looks like the query should be changed similar to mem usage: sum(rate(container_cpu_usage_seconds_total{namespace="$namespace", pod_name=~"$pod",container_name!="", container_name!="POD"}[1m])) by (pod_name)

P.S. the same for the CPU node usage

SergeiSporyshev commented 4 years ago

Hi, @eug-maly In our last release we changed all queries for cpu/memory usage metrics Can you check this https://github.com/devopsprodigy/kubegraf/releases/tag/v1.3.0beta?

eug-maly commented 4 years ago

looks good .thanks