Merge pull request 'gpu功能初始化' (#1) from dev1 into main

Reviewed-on: #1

app/k8sManager/multiple_clusters.py

@@ -207,6 +207,10 @@ def new_cluster_install(params):
             # "files/nfs-provisioner-deploy.yaml":"/opt/nfs-provisioner-deploy.yaml",
             "files/nfs-rbac.yaml": "/opt/nfs-rbac.yaml",
             "files/nvidia-device-plugin.yml": "/opt/nvidia-device-plugin.yml",
+            "files/libnvidia-container-tools_1.17.8-1_amd64.deb": "/opt/libnvidia-container-tools_1.17.8-1_amd64.deb",
+            "files/libnvidia-container1_1.17.8-1_amd64.deb": "/opt/libnvidia-container1_1.17.8-1_amd64.deb",
+            "files/nvidia-container-toolkit_1.17.8-1_amd64.deb": "/opt/nvidia-container-toolkit_1.17.8-1_amd64.deb",
+            "files/nvidia-container-toolkit-base_1.17.8-1_amd64.deb": "/opt/nvidia-container-toolkit-base_1.17.8-1_amd64.deb",
             "script/k8s_uninstall.sh": "/opt/k8s_uninstall.sh",
             "script/import_images.sh": "/opt/import_images.sh",
             }

app/k8sManager/ssh_utils.py

@@ -186,7 +186,7 @@ def execute_sudo_command(ssh, command, password, real_time_log, sudo_timeout, us
             if time.time() - start_time > sudo_timeout:
                 raise Exception(f"等待sudo密码提示超时（{sudo_timeout}秒）: {sudo_cmd}")
             if not ready:
-                time.sleep(1.5)  # 避免CPU占用过高
+                time.sleep(0.5)  # 避免CPU占用过高
 
         # 如果没有收到密码提示但命令执行超时，可能是权限问题
         if not password_prompt:
@@ -218,7 +218,7 @@ def execute_sudo_command(ssh, command, password, real_time_log, sudo_timeout, us
                     error += line
                 if channel.exit_status_ready():
                     break
-                time.sleep(1.5)  # 避免CPU占用过高
+                time.sleep(0.1)  # 避免CPU占用过高
         else:
             # 非实时模式读取输出
             result += channel.recv(-1).decode(errors="replace") if channel.recv_ready() else ""

files/nvidia-device-plugin.yml

@@ -0,0 +1,49 @@
+# Copyright (c) 2019, NVIDIA CORPORATION.  All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+apiVersion: apps/v1
+kind: DaemonSet
+metadata:
+  name: nvidia-device-plugin-daemonset
+  namespace: kube-system
+spec:
+  selector:
+    matchLabels:
+      name: nvidia-device-plugin-ds
+  updateStrategy:
+    type: RollingUpdate
+  template:
+    metadata:
+      labels:
+        name: nvidia-device-plugin-ds
+    spec:
+      tolerations:
+      - key: nvidia.com/gpu
+        operator: Exists
+        effect: NoSchedule
+      priorityClassName: "system-node-critical"
+      containers:
+      - image: nvcr.io/nvidia/k8s-device-plugin:v0.13.0
+        name: nvidia-device-plugin-ctr
+        securityContext:
+          allowPrivilegeEscalation: false
+          capabilities:
+            drop: ["ALL"]
+        volumeMounts:
+          - name: device-plugin
+            mountPath: /var/lib/kubelet/device-plugins
+      volumes:
+        - name: device-plugin
+          hostPath:
+            path: /var/lib/kubelet/device-plugins

script/k8s_install.sh

@@ -309,16 +309,30 @@ echo "containerd配置初始纠正完成."
 
 echo "开始更新containerd配置以适配GPU实例"
 
-# 检查是否有 NVIDIA GPU
+# 检查 NVIDIA GPU
 if lspci | grep -i nvidia > /dev/null 2>&1; then
-    log_info "检测到NVIDIA GPU,开始配置nvidia-container-runtime..."
+    if [ "$1" == "worker" ]; then
+        log_info "检测到NVIDIA GPU,开始配置nvidia-container-runtime..."
 
-    dpkg -i /opt/*.deb || log_error "安装nvidia-container-runtime及其依赖失败!"
+        # 检查 .deb 文件是否存在
+        if [ ! "$(ls /opt/*.deb 2>/dev/null | wc -l)" -ge 1 ]; then
+          log_error "/opt/ 下没有 .deb 文件"
+          exit 1
+        fi
 
-    # 配置 containerd 支持 nvidia runtime
-    CONTAINERD_CONFIG="/etc/containerd/config.toml"
-    if ! grep -q '\[plugins."io.containerd.grpc.v1.cri".containerd.runtimes.nvidia\]' "$CONTAINERD_CONFIG"; then
-        cat <<EOF >> "$CONTAINERD_CONFIG"
+        # 安装 .deb 包
+        for deb in /opt/*.deb; do
+          dpkg -i "$deb" || log_error "安装 $deb 失败"
+        done
+
+        # 配置 containerd
+        CONTAINERD_CONFIG="/etc/containerd/config.toml"
+        if ! grep -q '
+$$
+plugins."io.containerd.grpc.v1.cri".containerd.runtimes.nvidia
+$$
+' "$CONTAINERD_CONFIG"; then
+            cat <<EOF >> "$CONTAINERD_CONFIG"
 
 [plugins."io.containerd.grpc.v1.cri".containerd.runtimes.nvidia]
   privileged_without_host_devices = false
@@ -326,11 +340,17 @@ if lspci | grep -i nvidia > /dev/null 2>&1; then
   [plugins."io.containerd.grpc.v1.cri".containerd.runtimes.nvidia.options]
     BinaryName = "/usr/bin/nvidia-container-runtime"
 EOF
-    fi
+        fi
 
-    # 重启 containerd
-    systemctl restart containerd
-    log_info "nvidia-container-runtime 配置完成,containerd已重启"
+        # 设置 default_runtime_name = "nvidia"
+        if ! grep -q '^default_runtime_name = "nvidia"$' "$CONTAINERD_CONFIG"; then
+          sed -i '/$$plugins."io.containerd.grpc.v1.cri"$$/{n;s/.*/  default_runtime_name = "nvidia"/;}' "$CONTAINERD_CONFIG"
+        fi
+
+        # 重启 containerd
+        systemctl restart containerd
+        log_info "nvidia-container-runtime 配置完成,containerd已重启"
+    fi
 else
     log_info "未检测到NVIDIA GPU,跳过nvidia-container-runtime配置"
 fi