[Feature] Add nodepool-specific metrics for Cluster Autoscaler
- Dominant language
- TypeScript
- Stars
- 2.1k
- Forks
- 395
- Avg merge
- 2d 22h
- Merged PRs (30d)
- 13
Description
**Is your feature request related to a problem? Please describe.**
While troubleshooting some CAS issues, a customer who uses Azure Managed Prometheus noticed that the following metrics seem to be missing:
```
cluster_autoscaler_node_group_backoff_status
cluster_autoscaler_node_group_healthiness
```
They see these in the CAS source code for the version which should be present with AKS & k8s 1.33, and are able to access other metrics such as `cluster_autoscaler_nodes_count`.
**Describe the solution you'd like**
AKS autoscale engineers have determined that these two metrics were not showing up because we do not set a flag on our managed deployment:
```
emitPerNodeGroupMetrics = flag.Bool("emit-per-nodegroup-metrics", false, "If true, emit per node group metrics.")
```
* code ref for where this is used + which metrics are impacted: [https://github.com/kubernetes/autoscaler/blob/9952269641c8cb71f197e61bc3e74eb9582d572f/cluster-auto…](https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fkubernetes%2Fautoscaler%2Fblob%2F9952269641c8cb71f197e61bc3e74eb9582d572f%2Fcluster-autoscaler%2Fmetrics%2Fmetrics.go%23L467-L474&data=05%7C02%7Crgregory%40microsoft.com%7C6d041392cc0e40159f5d08de3dcc7b4f%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C639016145964963798%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=BYffTQUwMg4dNYNkr0kP5xjofQbJIqZ6PY%2FjsKr42k0%3D&reserved=0)
**Additional context**
This feature request is currently in our backlog. Please upvote this is you would like us to prioritize this workstream. Thanks.
Contributor guide
Assessment
This issue has not been assessed yet.