From c4582c99bcc5b3a3ad49cfb654f708e0bb1afe3a Mon Sep 17 00:00:00 2001
From: "sundapeng.sdp" <sundapeng@hashdata.cn>
Date: Mon, 20 Oct 2025 15:30:07 +0800
Subject: [PATCH] =?UTF-8?q?feat:=20metric=20e2e=20=E5=90=AF=E5=8A=A8?=
 =?UTF-8?q?=E6=9C=8D=E5=8A=A1=E4=B8=AD=E5=88=A4=E6=96=AD=E6=98=AF=E5=90=A6?=
 =?UTF-8?q?=E5=BD=93=E5=89=8D=E6=9C=8D=E5=8A=A1=E5=99=A8=E5=AD=98=E5=9C=A8?=
 =?UTF-8?q?=E5=8F=AF=E7=94=A8GPU=EF=BC=8C=E5=A6=82=E6=97=A0=E5=88=99?=
 =?UTF-8?q?=E8=B7=B3=E8=BF=87=20test-gpu-node=20=E5=AE=B9=E5=99=A8?=
 =?UTF-8?q?=E5=88=9B=E5=BB=BA=E5=92=8C=E5=AE=89=E8=A3=85=E5=8C=85=E7=9A=84?=
 =?UTF-8?q?=E6=B5=8B=E8=AF=95=EF=BC=9B?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

refs #29
---
 .../tests/scripts/04_test_gpu_node_install.sh | 28 ++++++---
 src/metric/tests/scripts/common/check-gpu.sh  | 59 +++++++++++++++++++
 src/metric/tests/scripts/common/start-all.sh  | 43 +++++++++++---
 3 files changed, 116 insertions(+), 14 deletions(-)
 create mode 100755 src/metric/tests/scripts/common/check-gpu.sh

diff --git a/src/metric/tests/scripts/04_test_gpu_node_install.sh b/src/metric/tests/scripts/04_test_gpu_node_install.sh
index ce1d19a..b0e2355 100755
--- a/src/metric/tests/scripts/04_test_gpu_node_install.sh
+++ b/src/metric/tests/scripts/04_test_gpu_node_install.sh
@@ -1,6 +1,9 @@
 #!/bin/bash
 set -e
 
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+COMMON_DIR="$SCRIPT_DIR/common"
+
 FTP_SERVER="${FTP_SERVER:-172.30.0.40}"
 FTP_USER="${FTP_USER:-ftpuser}"
 FTP_PASSWORD="${FTP_PASSWORD:-ZGClab1234!}"
@@ -8,26 +11,37 @@ FTP_PORT="${FTP_PORT:-21}"
 
 FTP_HOST="${FTP_SERVER}"
 
-echo "[03] 进入测试节点执行安装..."
-echo "[03] 使用 FTP 地址: ${FTP_HOST}:${FTP_PORT}"
+echo "[04] 检测GPU环境..."
+# 检测GPU环境
+if bash "$COMMON_DIR/check-gpu.sh"; then
+    echo "[04] GPU环境可用，继续执行GPU节点安装"
+    GPU_AVAILABLE=true
+else
+    echo "[04] GPU环境不可用，跳过GPU节点安装"
+    GPU_AVAILABLE=false
+    exit 0
+fi
+
+echo "[04] 进入测试节点执行安装..."
+echo "[04] 使用 FTP 地址: ${FTP_HOST}:${FTP_PORT}"
 
 docker exec argus-metric-test-gpu-node bash -c "
 set -e
 
 if ! command -v curl &>/dev/null; then
-    echo '[03] curl 未安装，正在安装...'
+    echo '[04] curl 未安装，正在安装...'
     apt-get update && apt-get install -y curl
 fi
 
 cd /tmp
-echo '[03] 下载 setup.sh...'
+echo '[04] 下载 setup.sh...'
 curl -u ${FTP_USER}:${FTP_PASSWORD} ftp://${FTP_HOST}:${FTP_PORT}/setup.sh -o setup.sh
 
-echo '[03] 执行安装...'
+echo '[04] 执行安装...'
 chmod +x setup.sh
 bash setup.sh --server ${FTP_HOST} --user ${FTP_USER} --password '${FTP_PASSWORD}' --port ${FTP_PORT}
 
-echo '[03] 安装完成'
+echo '[04] 安装完成'
 "
 
-echo "[03] 完成"
+echo "[04] 完成"
diff --git a/src/metric/tests/scripts/common/check-gpu.sh b/src/metric/tests/scripts/common/check-gpu.sh
new file mode 100755
index 0000000..c602304
--- /dev/null
+++ b/src/metric/tests/scripts/common/check-gpu.sh
@@ -0,0 +1,59 @@
+#!/bin/bash
+
+# GPU环境检测脚本
+# 检测系统是否有NVIDIA GPU硬件
+
+set -e
+
+# 检测函数
+check_gpu_support() {
+    echo "检测GPU环境..."
+    
+    # 方法1: 检测GPU设备文件
+    if ls /dev/nvidia* &>/dev/null; then
+        echo "✓ 检测到NVIDIA GPU设备文件"
+        return 0
+    fi
+    
+    # 方法2: 检测lspci中的NVIDIA设备（Linux）
+    if command -v lspci &> /dev/null; then
+        if lspci | grep -i nvidia &> /dev/null; then
+            echo "✓ 检测到NVIDIA GPU硬件"
+            return 0
+        fi
+    fi
+    
+    # 方法3: 检测nvidia-smi
+    if command -v nvidia-smi &> /dev/null; then
+        if nvidia-smi &> /dev/null; then
+            echo "✓ 检测到NVIDIA GPU硬件"
+            return 0
+        fi
+    fi
+    
+    echo "✗ 未检测到NVIDIA GPU硬件"
+    return 1
+}
+
+# 主函数
+main() {
+    echo "=========================================="
+    echo "  GPU环境检测"
+    echo "=========================================="
+    echo ""
+    
+    if check_gpu_support; then
+        echo ""
+        echo "结果: GPU环境可用"
+        exit 0
+    else
+        echo ""
+        echo "结果: GPU环境不可用，将跳过GPU相关服务"
+        exit 1
+    fi
+}
+
+# 如果直接运行此脚本
+if [ "${BASH_SOURCE[0]}" = "${0}" ]; then
+    main "$@"
+fi
diff --git a/src/metric/tests/scripts/common/start-all.sh b/src/metric/tests/scripts/common/start-all.sh
index 34d68df..7f0e7d5 100755
--- a/src/metric/tests/scripts/common/start-all.sh
+++ b/src/metric/tests/scripts/common/start-all.sh
@@ -59,18 +59,40 @@ echo "1. 初始化目录结构..."
 bash "$SCRIPT_DIR/init-directories.sh"
 
 echo ""
-echo "2. 检查 Docker 镜像..."
+echo "2. 检测GPU环境..."
+# 检测GPU环境
+if bash "$SCRIPT_DIR/check-gpu.sh"; then
+    echo "GPU环境可用，将启动GPU节点"
+    GPU_AVAILABLE=true
+else
+    echo "GPU环境不可用，跳过GPU节点"
+    GPU_AVAILABLE=false
+fi
+
+echo ""
+echo "3. 检查 Docker 镜像..."
 
 # 检查必要的镜像是否存在
-IMAGES=("argus-metric-ftp:latest" "argus-metric-prometheus:latest" "argus-metric-grafana:latest" "argus-metric-test-node:latest" "argus-metric-test-gpu-node:latest")
-missing_images=()
+BASE_IMAGES=("argus-metric-ftp:latest" "argus-metric-prometheus:latest" "argus-metric-grafana:latest" "argus-metric-test-node:latest")
+GPU_IMAGES=("argus-metric-test-gpu-node:latest")
 
-for image in "${IMAGES[@]}"; do
+# 先检查基础镜像
+missing_images=()
+for image in "${BASE_IMAGES[@]}"; do
     if ! docker images --format "{{.Repository}}:{{.Tag}}" | grep -q "^${image}$"; then
         missing_images+=("$image")
     fi
 done
 
+# 检查GPU镜像（如果GPU环境可用）
+if [ "$GPU_AVAILABLE" = true ]; then
+    for image in "${GPU_IMAGES[@]}"; do
+        if ! docker images --format "{{.Repository}}:{{.Tag}}" | grep -q "^${image}$"; then
+            missing_images+=("$image")
+        fi
+    done
+fi
+
 if [ ${#missing_images[@]} -gt 0 ]; then
     echo "以下镜像缺失，请先运行 build/build_images.sh 构建镜像："
     for image in "${missing_images[@]}"; do
@@ -85,10 +107,17 @@ else
 fi
 
 echo ""
-echo "3. 启动基础服务..."
+echo "4. 启动基础服务..."
 cd "$TEST_DIR"
-# 启动除GPU节点外的所有服务
-docker compose up -d ftp prometheus grafana test-node test-gpu-node
+
+# 根据GPU环境决定启动的服务
+if [ "$GPU_AVAILABLE" = true ]; then
+    echo "启动所有服务（包括GPU节点）..."
+    docker compose up -d ftp prometheus grafana test-node test-gpu-node
+else
+    echo "启动基础服务（跳过GPU节点）..."
+    docker compose up -d ftp prometheus grafana test-node
+fi
 
 echo ""
 echo "4. 等待服务启动..."