Metrics are just available for nodes and kube-system namespace

[Al4DIN]

Hello @serathius , @yangjunmyfm192085
I've the same issue on kubectl v1.22.5 client & v1.22.1 for the server, I can only retrieve metrics from node and some pods from my kube-system since I installed my metrics-server on this namespace but can't get other pods metrics.
I'm using k8s.gcr.io/metrics-server/metrics-server:v0.4.1
This is the result of my kubectl get --raw /api/v1/nodes/minikube2/proxy/metrics/resource

# HELP node_cpu_usage_seconds_total [ALPHA] Cumulative cpu time consumed by the node in core-seconds
# TYPE node_cpu_usage_seconds_total counter
node_cpu_usage_seconds_total 1762.835513 1651568930091
# HELP node_memory_working_set_bytes [ALPHA] Current working set of the node in bytes
# TYPE node_memory_working_set_bytes gauge
node_memory_working_set_bytes 3.749752832e+09 1651568930091
# HELP scrape_error [ALPHA] 1 if there was an error while getting container metrics, 0 otherwise
# TYPE scrape_error gauge
scrape_error 0

Originally posted by @Al4DIN in #976 (comment)

[yangjunmyfm192085]

Hi @Al4DIN, metrics-server:v0.4.1 use data from the kubelet stats/summary endpoint, not metrics/resource.

you can try to use

NODE_NAME=<Name of node in your cluster>
kubectl get --raw /api/v1/nodes/$NODE_NAME/proxy/stats/summary | jq '{cpu: .node.cpu, memory: .node.memory}'

to get raw data.

[yangjunmyfm192085]

You can check if it exists #976 (comment) mentioned in the case.

[Al4DIN]

Thank you for your reponse, here's my command result, and since I'm using minikube on Windows
kubectl get --raw /api/v1/nodes/$NODE_NAME/proxy/stats/summary

{
 "node": {
  "nodeName": "minikube2",
  "systemContainers": [
   {
    "name": "kubelet",
    "startTime": "2022-05-03T08:26:18Z",
    "cpu": {
     "time": "2022-05-04T10:11:11Z",
     "usageNanoCores": 0,
     "usageCoreNanoSeconds": 23102720953300
    },
    "memory": {
     "time": "2022-05-04T10:11:11Z",
     "availableBytes": 78282752,
     "usageBytes": 4188815360,
     "workingSetBytes": 4116021248,
     "rssBytes": 3816587264,
     "pageFaults": 40227,
     "majorPageFaults": 99
    }
   },
   {
    "name": "runtime",
    "startTime": "2022-05-03T08:25:53Z",
    "cpu": {
     "time": "2022-05-04T10:11:11Z",
     "usageNanoCores": 0,
     "usageCoreNanoSeconds": 23102720953300
    },
    "memory": {
     "time": "2022-05-04T10:11:11Z",
     "availableBytes": 78282752,
     "usageBytes": 4188815360,
     "workingSetBytes": 4116021248,
     "rssBytes": 3816587264,
     "pageFaults": 40227,
     "majorPageFaults": 99
    }
   },
   {
    "name": "pods",
    "startTime": "2022-05-03T08:26:19Z",
    "cpu": {
     "time": "2022-05-04T10:12:02Z",
     "usageNanoCores": 490859521,
     "usageCoreNanoSeconds": 15090179481700
    },
    "memory": {
     "time": "2022-05-04T10:12:02Z",
     "availableBytes": 10298327040,
     "usageBytes": 3068940288,
     "workingSetBytes": 2997530624,
     "rssBytes": 2794213376,
     "pageFaults": 0,
     "majorPageFaults": 0
    }
   }
  ],
  "startTime": "2022-05-02T20:33:00Z",
  "cpu": {
   "time": "2022-05-04T10:12:00Z",
   "usageNanoCores": 1001610985,
   "usageCoreNanoSeconds": 23143142465300
  },
  "memory": {
   "time": "2022-05-04T10:12:00Z",
   "availableBytes": 9185341440,
   "usageBytes": 4188213248,
   "workingSetBytes": 4110516224,
   "rssBytes": 3810488320,
   "pageFaults": 40227,
   "majorPageFaults": 99
  },
  "network": {
   "time": "2022-05-04T10:12:00Z",
   "name": "eth0",
   "rxBytes": 327792110,
   "rxErrors": 0,
   "txBytes": 131548915,
   "txErrors": 0,
   "interfaces": [
    {
     "name": "tunl0",
     "rxBytes": 0,
     "rxErrors": 0,
     "txBytes": 0,
     "txErrors": 0
    },
    {
     "name": "sit0",
     "rxBytes": 0,
     "rxErrors": 0,
     "txBytes": 0,
     "txErrors": 0
    },
    {
     "name": "eth0",
     "rxBytes": 327792110,
     "rxErrors": 0,
     "txBytes": 131548915,
     "txErrors": 0
    }
   ]
  },
  "fs": {
   "time": "2022-05-04T10:12:00Z",
   "availableBytes": 188654366720,
   "capacityBytes": 269490393088,
   "usedBytes": 67075354624,
   "inodesFree": 15624261,
   "inodes": 16777216,
   "inodesUsed": 1152955
  },
  "runtime": {
   "imageFs": {
    "time": "2022-05-04T10:12:00Z",
    "availableBytes": 188654366720,
    "capacityBytes": 269490393088,
    "usedBytes": 14249286657,
    "inodesFree": 15624261,
    "inodes": 16777216,
    "inodesUsed": 1152955
   }
  },
  "rlimit": {
   "time": "2022-05-04T10:12:05Z",
   "maxpid": 4194304,
   "curproc": 1974
  }
 },
 "pods": []
} 

We can clearly see that for pods I'm not geting any ressource. I've also tried many solution on stackoverflow and github issues but still can't figure out the cause.

[yangjunmyfm192085]

Yeah, Your analysis is correct.
Then this is not a metrics-server issue.
It is recommended to open an issue in kubernetes/kubernetes and try to analyze this issue.

[yangjunmyfm192085]

/kind support

[Al4DIN]

Okay, thank you.
I'm closing this issue.

[Al4DIN]

Actually It was relate to Minikube Environnement where you need to add an argument while launching your minikube minikube start --extra-config=kubelet.housekeeping-interval=10s

[serathius]

Nice, looks like something we could document as a known issue.

[Al4DIN]

The problem is that an old issue that It's still not pushed in the release version of Minikube.

[serathius]

I mean yes, it would be ideal for owners of K8s distribution to fix/document their distro. From Metrics Server side we only can ask Minikube nicely and document it ourselves.

[yangjunmyfm192085]

Yeah, we should document it as a known issue and I'll update the known issue.

[marc-wilson]

Is there an equivalent setting for docker desktop? Having the same issue.

[hecheng64]

Yeah, we should document it as a known issue and I'll update the known issue.

How to solve this problem