Unable to use g4dn instance types (follow on from https://github.com/aws/karpenter-provider-aws/issues/5676)

[andrewhibbert]

Description

Observed Behavior:

Follow on from #5676. Ideally shouldn't have to change settings.vmMemoryOverheadPercentage should work with similar instance types

{"error":"incompatible with nodepool \"gpu\", daemonset overhead={\"cpu\":\"690m\",\"memory\":\"843162368\",\"pods\":\"11\"}, no instance type satisfied resources {\"cpu\":\"1690m\",\"memory\":\"1916904192\",\"nvidia.com/gpu\":\"1\",\"pods\":\"12\"} and requirements karpenter.k8s.aws/ec2nodeclass In [gpu], karpenter.k8s.aws/instance-family In [g4ad] minValues 1, karpenter.k8s.aws/instance-size In [2xlarge] minValues 1, karpenter.sh/capacity-type In [on-demand spot], karpenter.sh/nodepool In [gpu], kubernetes.io/arch In [amd64], kubernetes.io/os In [linux], nvidia.com/gpu In [true] (no instance type has enough resources)"},{"error":"incompatible with nodepool \"enrichment-service\", daemonset overhead={\"cpu\":\"690m\",\"memory\":\"843162368\",\"pods\":\"10\"}

This works fine with a g5.2xlarge however.

Expected Behavior:

Reproduction Steps (Please include YAML):

Versions:
1.2.0

• Chart Version:
• Kubernetes Version (kubectl version):

• Please vote on this issue by adding a 👍 reaction to the original issue to help the community and maintainers prioritize this request
• Please do not leave "+1" or "me too" comments, they generate extra noise for issue followers and do not help prioritize the request
• If you are interested in working on this issue or have submitted a pull request, please leave a comment