Open universam1 opened 2 hours ago
This issue is currently awaiting triage.
If Karpenter contributors determines this is a relevant issue, they will accept it by applying the triage/accepted
label and provide further guidance.
The triage/accepted
label can be added by org members by writing /triage accepted
in a comment.
Description
Observed Behavior:
When using Admission controller: ExtendedResourceToleration functionality it causes Karpenter to fail to schedule nodes. It is defined via a resource request which will trigger the Admission controller to create a Toleration for the matching taint.
However, Karpenter presumptuously believes it should decide to fail as it "cannot provide that resources".
Expected Behavior:
Karpenter to ignore unknown resources, the same way as ClusterAutoscaler does!
Reproduction Steps (Please include YAML):
Follow the steps in https://medium.com/robovision-ai/use-kubernetes-extendedresourcetoleration-to-keep-out-non-gpu-pods-c180f5c5e76d