菜单
Manage SR-IOV virtual functions, PCI devices, link speed, and RDMA testing for NVIDIA Network Operator. Use when setting VFs, querying PCI devices, forcing link speed, running RDMA bandwidth tests, checking RDMA capability, managing node labels and taints, or cordoning nodes.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Configure netop-tools for a specific hardware platform and use case. Use when setting up global_ops_user.cfg, choosing a use case (sriovnet_rdma, sriovibnet_rdma, hostdev, ipoib, macvlan), selecting IPAM type, configuring NETOP_NETLIST device list, setting feature flags, or switching between platforms (DGX, Dell, Lenovo, OCI, SuperMicro).
Deploy NVIDIA Network Operator to a Kubernetes cluster via netop-tools. Use when installing network operator, setting up RDMA networking, deploying SR-IOV, running ins-network-operator.sh, generating Helm values, applying network CRDs, or troubleshooting deployment failures.
Run and create tests for netop-tools configuration generation. Use when running unitest.sh, validating YAML generation, creating new test cases, checking test baselines, debugging test failures, or verifying config changes don't break existing tests.
Diagnose and fix NVIDIA Network Operator issues in Kubernetes. Use when pods can't get RDMA network, SR-IOV VFs not showing, IP allocation failures, network attachment missing, must-gather diagnostics, operator pods crashing, NicClusterPolicy errors, or IPAM pool exhaustion.
Upgrade NVIDIA Network Operator to a new version. Use when changing NETOP_VERSION, running upgrade-network-operator.sh, migrating between operator versions, handling CRD schema changes, or rolling back a failed upgrade.
| name | managing-netop-devices |
| description | Manage SR-IOV virtual functions, PCI devices, link speed, and RDMA testing for NVIDIA Network Operator. Use when setting VFs, querying PCI devices, forcing link speed, running RDMA bandwidth tests, checking RDMA capability, managing node labels and taints, or cordoning nodes. |
Device management, RDMA testing, and node operations for Network Operator clusters.
# Set VFs on PCI devices (run on worker node)
./setvfs.sh <NUM_VFS> <BDF1> [BDF2] ...
./setvfs.sh 8 0000:08:00.0 0000:86:00.1
# Query current VF count
./ops/getnumvfs.sh
# Python CLI
python3 python_tools/netop_tools.py ops device set-vfs 8
python3 python_tools/netop_tools.py ops device get-vfs
./ops/getpci.sh # List PCI devices
./ops/getpciid.sh # Vendor IDs (15b3 = NVIDIA/Mellanox)
./ops/grabpci.sh # PCI config details
./ops/devlist.sh # Network devices
# Force link speed (disables auto-negotiation)
./ops/force_link_speed.sh <DEV> <SPEED> <XVAL>
# Speeds: 10G, 25G, 40G, 50G, 100G, 200G, 400G, 800G
# XVAL: 1X, 2X, 4X, 8X (must match speed)
./ops/force_link_speed.sh mlx5_0 400G 4X
# Check current speed
./ops/chklnkspeed.sh
./ops/linkchk.sh
# Verify RDMA capability
./rdmatest/check_rdma.sh
./rdmatest/get_rdma_dev.sh
# Disable PCI ACS for peer-to-peer (bare metal only)
./rdmatest/disable_acs.sh
# RoCE bandwidth test (server → client)
./rdmatest/rocesrv.sh # On server pod
./rdmatest/roceclnt.sh # On client pod
# InfiniBand bandwidth test
./rdmatest/rdmasrv.sh # Server: ib_send_bw
./rdmatest/rdmaclnt.sh # Client
# GPU Direct RDMA
./rdmatest/gdrsrv.sh # Server
./rdmatest/gdrclt.sh # Client
# Performance testing with CUDA
./rdmatools/perftestcuda.sh
# Labeling
./ops/labelworker.sh # Label workers
./ops/labelmaster.sh # Label control plane
./ops/labelsu.sh # Label for scalable units
./ops/dellabelworker.sh # Remove worker labels
# Taints
./ops/gettaints.sh # Show taints
./ops/rmtaints.sh # Remove NoSchedule taints
# Cordon (sourced as functions)
source ${NETOP_ROOT_DIR}/ops/cordon.sh
cordon # Cordon all workers
uncordon # Uncordon all workers
# Python CLI
python3 python_tools/netop_tools.py ops node cordon
python3 python_tools/netop_tools.py ops node uncordon
python3 python_tools/netop_tools.py ops node label rdma-capable true
# Control plane as worker
./ops/add-controlplane-as-worker.sh <NODENAME>
./ops/rm-controlplane-as-worker.sh <NODENAME>
# Create test pod YAML
${NETOP_ROOT_DIR}/ops/mk-app.sh <name> [num_pods] [namespace] [worker_node]
${NETOP_ROOT_DIR}/ops/mk-app.sh test 2 default worker-01
# Deploy
${NETOP_ROOT_DIR}/ops/run-app.sh test