[Exoscale]: Instance pools always default back to 1 min size

[francoismeulenberg]

Which component are you using?:
/area cluster-autoscaler

What version of the component are you using?:
1.35.0

Component version: cluster-autoscaler-chart-9.56.0

What k8s version are you using (kubectl version)?:

kubectl version Output

$ kubectl version 
Server Version: v1.34.4+rke2r1

What environment is this in?:
Exoscale

What did you expect to happen?:
I would expect the cluster autoscaler to respect the minimum set instances on a Instance Pool.

What happened instead?:
The cluster autoscaler ignores the minimum set instances and always defaults back onto 1

How to reproduce it (as minimally and precisely as possible):

1. Deploy 3 static controlplane instances on Exoscale
2. Deploy a self-managed kubernetes cluster, no SKS, on Exoscale with cloud provider set to external.
3. Deploy the exoscale Cloud Controller Manager.
4. Deploy a Instance pool with a mimum set to three to be configured as worker nodes.
5. Deploy the Cluster Autoscaler with the Exoscale configuration.
6. Wait for the CA to be fully active and see that it automatically scales back to 1

Anything else we need to know?:
Based on my analysis on these two files:

• exoscale_node_group_instance_pool.go
• exoscale_node_group_sks_nodepool.go

The exoscale_node_group_sks_nodepool.go has parameters for setting the Instance Pool size but the exoscale_node_group_instance_pool.go does not which is why the CA seems to default back 1 as it's hardcoded.

[Debasish-87]

It looks like exoscale_node_group_instance_pool.go does not propagate or respect the configured minimum size, whereas exoscale_node_group_sks_nodepool.go explicitly handles it.

From the code path, this seems to result in a fallback to a default value (1), which matches the observed behavior.

Would it make sense to align the instance pool implementation with the SKS nodepool logic so that the configured min size is consistently respected?

[francoismeulenberg]

yeah, I would think so.