Azure / AKS

Azure Kubernetes Service
https://azure.github.io/AKS/
1.95k stars 304 forks source link

[Feature] scraping Cluster Autoscaler metrics to unmanaged Prometheus #3877

Open otfung opened 1 year ago

otfung commented 1 year ago

Is your feature request related to a problem? Please describe. I'm using Cluster Autoscaler and want to know how to get detailed metrics like all metrics for cluster autoscaling, current cluster autoscaling values, max, min, etc. exported via Prometheus. In essence, what I want is to get metrics of % of max nodepool scaling count. Is there a Prometheus endpoint or somewhere on the AKS cluster we can scrape to get cluster autoscaler metrics from data?

I opened a support request (2308140040000237) with MS support and got to know that those metrics are provided by CAS. They shared what metrics the upstream project believes CAS should support: autoscaler/cluster-autoscaler/proposals/metrics.md at master · kubernetes/autoscaler · GitHub

Later he concluded that I don't think that's going to hit what the customer is wanting. scraping Cluster Autoscaler metrics to unmanaged Prometheus

Describe alternatives you've considered N/A

aritraghosh commented 1 year ago

@otfung We do not have autoscaler metrics available for scraping through Prometheus right now. There are subset of the autoscaler metrics available through platform metrics. In addition, we are working on enabling customers to get autoscaler metrics through Azure Managed Prometheus, which we expect to have by Dec 2023.

grzesuav commented 7 months ago

https://learn.microsoft.com/en-us/azure/aks/monitor-control-plane-metrics there is something regarding cluster autoscaler