Worker pool priorities do not get respected

[tobschli]

How to categorize this issue?

/area quality
/kind bug

What happened:

When specifying the priorities of worker pools, introduced in #11045, they do not get respected, as the cluster-autoscaler does not accept the generated yaml:

W0131 17:51:12.980662       1 priority.go:90] Wrong configuration for priority expander: Can't parse YAML with priorities in the configmap: yaml: unmarshal errors:                                                                                                                   │
line 1: cannot unmarshal !!str `0` into int                                                                                                                                                                                                                                         │
line 3: cannot unmarshal !!str `100` into int. Ignoring update.

What you expected to happen:

The cluster-autoscaler should respect the generated yaml and use them respectively.

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

1. Create a shoot with at least two worker groups, from which at least one has a priority defined.
2. Observe the cluster-autoscaler logs.

Anything else we need to know?:

Note

Although this bug exists, even when only the priority expander is specified, the cluster can scale the worker groups. The priorities are simply ignored.

Environment:

• Gardener version: v1.112.0-dev-c26b56e05e6ae7211290bd8b6b0146dadd46c6a2
• Kubernetes version (use kubectl version): 1.131.4
• Cloud provider or hardware configuration: aws
• Others:

[tobschli]

/assign

[tobschli]

The issue probably lies in the fact that the kubernetes yaml Marshaller encloses the priorities with double-quotes, even though the struct has int32's as keys.

In the example of cluster-autoscaler does not have these quotes.