timeout for node registration

[saheerb]

Hi,

I’m using Karpenter v0.37.5 (from oci://public.ecr.aws/karpenter/karpenter) in a CI environment where bursts of jobs create many pods at once.

When a CI job is scheduled, Karpenter correctly spins up a new node in AWS.
However, sometimes node registration (the kubelet joining the cluster) takes a few seconds. During that short delay, Karpenter seems to think capacity is still insufficient and launches additional nodes, even though the first node is still booting and will register shortly after.

Is there any variable or configuration setting to control how long Karpenter should wait for node registration before deciding to launch more nodes?
Basically, a way to tune the “node registration timeout” or grace period before triggering new provisioning?

Thanks for your help!