PriorityClass Setting Fails Due to Deprecation of scheduling.k8s.io/v1beta1 API in K8s 1.22 onwards

[morampudisouji]

trafficstars

Bug report:

The PriorityClass setting is not working as expected because the cluster is running EKS version 1.28, where the scheduling.k8s.io/v1beta1 API version has been deprecated. Currently, the chart is checking for scheduling.k8s.io/v1beta1, which is causing issues. This problem is evident in the Helm chart file at client-daemonset.yaml#L63.

As of Kubernetes v1.22, the scheduling.k8s.io/v1beta1 API version for PriorityClass is no longer served. Consequently, the priorityclassname configuration is not being applied correctly.

Reference:

For more information, refer to the Kubernetes deprecation guide.

Expected behavior:

The priorityclassname setting should be applied correctly, and the appropriate PriorityClass should be created or utilized based on the current API version supported by the cluster.

How to reproduce it:

Check the applied PriorityClass settings in the cluster with eks version 1.28. Observe that instead of the specified priorityclassname, the default priority class, "normal," is applied.

Environment:

• Dragonfly version: I tried in helm chart 1.1.61 and 1.2.1 versions
• OS:
• Kernel (e.g. uname -a): 5.10.218-208.862.amzn2.x86_64 #1 SMP Tue Jun 4 16:52:10 UTC 2024 x86_64 x86_64 x86_64 GNU/Linux
• Others: