Does anyone have Prometheus metrics integration working with helm-based installs?

[shadowzen1978]

This is an issue that started happening for me some time back with OpenLens and continues in FreeLens.

I use the standard prometheus-community helm chart to install Prometheus into a monitoring namespace in each of my clusters. This helm chart sets up the service name as prometheus-server.

The service address of Prometheus per the format in the cluster settings should, therefore, be monitoring/prometheus-server:9090 by my reckoning, but metrics don't work for me. I've tried with Helm and Helm 14.x formats (not sure what the difference here is, btw).

I would love to get this working, if possible, and I've searched online for solutions with no luck. I feel like this is either broken or something stupidly simple that I am overlooking.

[dex4er]

I use kube-prometheus-stack from https://prometheus-community.github.io/helm-charts and it works perfectly with "Auto Detect Prometheus" mode.

[robertobandini]

Hi @shadowzen1978,

I can confirm that with the old version you need to specify the Prometheus internal endpoint in that case, meanwhile, as @dex4er said, with kube-prometheus auto detect works.

I already tought that we have to verify and fix.
@shadowzen1978 if you can confirm this problem is still present also on the last Freelens version, then please open an issue about it if nobody have already did.

Thanks for your report @shadowzen1978!

[nowll]

I encountered the same issue while configuring Prometheus.
I tried installing it using the official Helm chart from [prometheus-community.github.io/helm-charts]| , but it didn’t work — regardless of whether I used Helm version 14.x, Helm or the auto-detect option.

Here’s the command I ran:
helm install prometheus prometheus-community/kube-prometheus-stack -n monitoring --create-namespace

Did I do something wrong?
Also, should I expose or forward the service using a NodePort instead?

[robertobandini]

As a feedback I can say:

Recently I often create a Kind Kubernetes cluster v1.33.1 on Ubuntu WSL2 and I add both metrics server and kube-prometheus-stack with:

helm upgrade --install metrics-server metrics-server/metrics-server

and

helm install kube-prometheus oci://ghcr.io/prometheus-community/charts/kube-prometheus-stack

Using the last Freelens release everything works fine

[takashoty]

Well it is no luck with setup 'helm install prometheus prometheus-community/kube-prometheus-stack -n monitoring'.

Same issue for me. Neither service kube-prometheus-stack-prometheus nor its endpoint didn't bring visible metrics for freelens.
Issue exists.

[robertobandini]

Hi @takashoty, could you please open an issue specifying the Freelens version, the version of the software used for the metrics, the OS, etc.? So we can try to replicate and resolve the issue together. If it's as you say, the problem is specific to some cases, as it works flawlessly in others. Thank you very much, I hope we can help you asap!

[alexvanderberkel]

@robertobandini 1.7.0 works for me, where 1.8.0 does not anymore. Running it on kube-prometheus-stack v. 81.4.2' on talos linux

[robertobandini]

Thanks @alexvanderberkel It may be as reported here: #1078 (comment)