i'm using the same label and the taints in the deployment manifests, but i still see the below error in karpenter logs

{"level":"ERROR","time":"2024-05-29T16:01:08.723Z","logger":"controller.provisioner","message":"Could not schedule pod, incompatible with nodepool "default", daemonset overhead={"cpu":"736m","memory":"1067557632","pods":"8"}, incompatible requirements, label "application" does not have known values","commit":"6b868db","pod":"default/rancher-test-98559869c-5vdw8"}
{"level":"ERROR"

Warning FailedScheduling 0s karpenter Failed to schedule pod, incompatible with nodepool "default", daemonset overhead={"cpu":"766m","memory":"1193386752","pods":"9"}, incompatible requirements, label "application" does not have known values

from karpenter.

failing to understand why is the karpenter failing to spin up a node when the labels and nodeselctors match and using the same tolerations in deployment file

from karpenter.

The log you provided has a commit indicating that it's from Karpenter v0.36.0 and has a different NodePool name than the Provisioner you provided. Also a Provisioner wouldn't be compatible with v0.36.0, dual support for the alpha and beta APIs was dropped at v0.33.0. Do you have a NodePool in your cluster that you would expect this pod to schedule to other than "default", which Karpenter considers incompatible?

from karpenter.

sorry, i didnt get you completely since i'm new to karpenter, i just have a single nodepool which is the default one
kubectl get nodepool
NAME NODECLASS
default default

from karpenter.

is there any difference between nodepool and provisioners? my understanding may be incorrect, i deployed karpenter as an addon, and using helm i will have to deploy multiple provisioners,

from karpenter.

{{- range .Values.provisioners }}

apiVersion: karpenter.sh/v1alpha5
kind: Provisioner
metadata:
name: {{ .name }}
spec:
labels:
{{- range $key, $value := .labels }}
{{ $key }}: {{ $value | quote }}
{{- end }}
requirements:
- key: "node.kubernetes.io/instance-type"
operator: In
values:
{{- range .instanceTypes }}
- {{ . }}
{{- end }}
- key: "karpenter.sh/capacity-type"
operator: In
values:
{{- range .capacityType }}
- {{ . }}
{{- end }}
limits:
resources:
cpu: {{ .limit }}
{{- if .taints }}
taints:
- effect: NoSchedule
key: {{ .taints.key | quote }}
value: {{ .taints.value | quote }}
{{- end }}
providerRef:
name: {{ .providerReference }}
ttlSecondsAfterEmpty: 30
{{- end }}

this is the provisioner templabe file i'm using

from karpenter.

i would like to have the pods scheduled on the nodes which are having the same config i pass through the provisioners and not the default nodepool

from karpenter.

@jmdeal , correct me if my understanding is wrong, for karpenter 0.36.x, we have karpenter node pools instead of karpenter provisioners? from the github repo i see nodepool repo instead of a provisioner repohttps://github.com/aws/karpenter-provider-aws/tree/v0.36.1/charts/karpenter/crds

from karpenter.

Provisioners replaced NodePools in Karpenter v0.32.0, and support for Provisioners was completely removed in v0.33.0. You'll need to replace your provisioners with equivalent nodepools for them to work with v0.36.x. You can find more information in the v1beta1 migration guide.

from karpenter.