Aims to fix https://github.com/stefanprodan/k8s-prom-hpa/issues/20

[stafot]

Based on feedback from AWS EKS team which told me:
" looks like you're running a pre 0.3.0 metrics server
You need 0.3.0, as it contains the latest K8s dependencies, which contain our fix"

[stefanprodan]

Hi @stafot thanks for the PR. Have you tested this on EKS?

[stafot]

@stefanprodan Tested today. made some follow up fixes/additions which were based on this approach. And now it is working.

If you want more info let me know.

[stefanprodan]

Thanks for testing it. With this config do you get the podinfo http_requests metric when you do kubectl get --raw "/apis/custom.metrics.k8s.io/v1beta1/namespaces/default/pods/*/http_requests" ?

[stafot]

@stefanprodan Nope. Probably I need a different config file based on my needs or a more generic one. Am I right?

The response is the following:

Error from server (NotFound): the server could not find the metric http_requests for pods

[stafot]

@stefanprodan Even if custom-metrics-apiserver looks to work as expected based on log outputs I get, None of the raw kubectl commands you shared for testing works. So probably I have something misconfigured, still don't find what it is though yet.

[stafot]

@stefanprodan I can confirm that this PR works as expected, so if you don;t mind please merge it. In my case the problem was that I used kube-prometheus so my prometheus url was prometheus-k8s and not prometheus as custom-metrics-apiserver expected by default. So a possible improvement to my PR could be to make prometheus url environment variable to be clearer that might varies in different use cases.

[stefanprodan]

Thanks @stafot