You are deploying the NVIDIA GPU Operator in a Kubernetes cluster to bootstrap your nodes for accelerated AI workflows. Which of the following components are automatically installed and managed by this operator on your GPU worker nodes? (Select all that apply.)
Select all correct answers, then click Submit.
Short Explanation
Pay close attention here: the NVIDIA GPU Operator automates the GPU-specific plumbing on your worker nodes. It deploys the NVIDIA Container Toolkit to connect containers to GPUs, the NVIDIA Device Plugin to expose GPUs to the scheduler, and the GPU Feature Discovery plugin to label node capabilities. CoreDNS, the container runtime, and the API server are standard Kubernetes pieces the operator uses but doesn't manage.
Full Explanation
The NVIDIA GPU Operator automates the lifecycle of NVIDIA driver and software components inside Kubernetes. Instead of manually installing GPU drivers, container runtime configurations, and monitoring tools on every node, the operator deploys them as containerized workloads. The components it automatically installs and manages on GPU worker nodes are the NVIDIA Container Toolkit, the NVIDIA Device Plugin, and the NVIDIA GPU Feature Discovery plugin. The NVIDIA Container Toolkit mounts the GPU driver libraries into the container at startup, enabling the containerized application to interface with the GPU. The NVIDIA Device Plugin exposes the GPU count as a schedulable resource, such as 'nvidia.com/gpu', so the Kubernetes scheduler can allocate pods to nodes with available GPUs. The NVIDIA GPU Feature Discovery plugin automates node labeling by inspecting GPU properties, such as memory capacity, CUDA driver version, and GPU model, and applying relevant labels to nodes, which allows workloads to be targeted using node affinity. Standard Kubernetes CoreDNS for cluster-wide name resolution, the container runtime engine hosting the pods, such as Docker or Containerd, and the Kubernetes API Server control plane daemon are not managed by the operator. They are standard cluster control plane or worker node prerequisites. The operator relies on these existing components but does not deploy or manage them.