Hi,

thanks a lot for the hint. It works fine under Yoga as you said:

$ kubectl get apiservice | grep metrics
v1beta1.metrics.k8s.io                 kube-system/magnum-metrics-server   True        9m11s
$ kubectl top nodes
NAME                             CPU(cores)   CPU%   MEMORY(bytes)   MEMORY%
k8s-test-dhgkjeop7o3v-master-0   139m         3%     1227Mi          15%
k8s-test-dhgkjeop7o3v-node-0     29m          1%     495Mi           12%
k8s-test-dhgkjeop7o3v-node-1     38m          1%     567Mi           14%

just out of curiosity, will this be fixed in e.g. Wallaby or Xena as well?

Cheers,

Oliver

Am 04.05.2022 um 09:04 schrieb Oliver Weinmann:
Hi ammad,

Thanks for your reply. I will try this.

Best regards

Von meinem iPhone gesendet

Am 04.05.2022 um 09:00 schrieb Ammad Syed <syedammad83@gmail.com>:


Hi,

You can use yoga release of magnum. Metrics server has fixed in it.

On the older release use metrics server enable label to false.

Ammad
On Tue, May 3, 2022 at 11:36 PM Oliver Weinmann <oliver.weinmann@me.com> wrote:
Hi,

Is the metrics-server broken by default in magnum? I tried for several hours to get it running, but running kubectl top nodes fails.

Message:               failing or missing response from https://10.100.2.4:8443/apis/metrics.k8s.io/v1beta1: Get "https://10.100.2.4:8443/apis/metrics.k8s.io/v1beta1": net/http: request canceled while waiting for connection (Client.Timeout exceeded while awaiting headers)


I also googled a lot but only found similar issues for k8s clusters not deployed via magnum. I remembered from my other cluster that I had to enable enable-aggregator-routing=true in kupe-apiserver and --kubelet-insecure-tls for the metrics-server.
But this also doesn’t help. Any help would be highly appreciated.

Best regards,
Oliver
--
Regards,


Syed Ammad Ali