How do CPU and memory requests work with GPU Pods?
GPU Pods request GPUs through the nvidia.com/gpu resource, but their CPU and memory requests behave like any other Kuber …
How do I request specific GPU types using Node selectors?
Use a nodeSelector (or nodeAffinity) on your Pod to target a specific GPU type. CoreWeave labels GPU Nodes with hardware …
How do I use Kueue for job queuing?
Install Kueue on your CKS cluster from the CoreWeave-published Helm chart (coreweave/cks-kueue), then define ResourceFla …
How do Kubernetes QoS classes work?
Kubernetes assigns each Pod one of three Quality-of-Service classes based on its resource requests and limits, and the k …
What are the recommended resource requests and limits for GPU Pods?
For most GPU training and inference workloads, set CPU and memory requests to roughly the per-GPU share of the instance, …
What is the autoscaler behavior when Pods are pending?
The CKS autoscaler scales a Node Pool up only when adding another Node of that pool’s instance type would make a pending …
Why are my Pods not scheduling on GPU Nodes?
GPU Pods stay Pending when they request a resource the available GPU Nodes do not provide, when no Node Pool of the requ …