GPU 申请
GPU 申请
resources:
limits:
nvidia.com/gpu: 1 # 扩展资源只写 limits 即可,requests 自动相等核对节点资源
核对节点资源
kubectl describe node gpu-node1 | grep -A8 'Capacity:' | grep -i gpu
kubectl -n kube-system get pods | grep -i deviceCPU/内存之外的设备资源(nvidia.com/gpu 等)由 Device Plugin 上报;申请时 requests 与 limits 必须相等且为整数。
resources:
limits:
nvidia.com/gpu: 1 # 扩展资源只写 limits 即可,requests 自动相等kubectl describe node gpu-node1 | grep -A8 'Capacity:' | grep -i gpu
kubectl -n kube-system get pods | grep -i device