CoreWeave Upgrade & Migration

GPU Type Migration

# Before: A100
affinity:
  nodeAffinity:
    requiredDuringSchedulingIgnoredDuringExecution:
      nodeSelectorTerms:
        - matchExpressions:
            - key: gpu.nvidia.com/class
              operator: In
              values: ["A100_PCIE_80GB"]

# After: H100 (update affinity label)
              values: ["H100_SXM5"]

Migration Steps

1. Build new container with updated CUDA version if needed
2. Deploy with new GPU affinity alongside old deployment
3. Shift traffic gradually (canary)
4. Validate latency and throughput match or improve
5. Scale down old deployment

CUDA Version Matrix

Rollback

kubectl rollout undo deployment/my-inference

Resources

• CoreWeave GPU Instances

Next Steps

For CI/CD, see coreweave-ci-integration.