feat(deploy): 增加受控更新与环境配置迁移

update.sh

@@ -15,6 +15,10 @@ if [[ "${1:-}" =~ ^(-h|--help)$ ]]; then
     echo "  UPDATE_SKIP_REPO_UPDATE=true|false"
     echo "      true  等同 UPDATE_MODE=image"
     echo "      false 等同 UPDATE_MODE=full"
+    echo "  UPDATE_HEALTH_TIMEOUT_SECONDS=秒数"
+    echo "      等待核心服务恢复可用的最长时间，默认 300 秒"
+    echo ""
+    echo "仓库更新后会自动执行 scripts/update-env.d 中的幂等环境配置迁移。"
     exit 0
 fi
 
@@ -100,8 +104,9 @@ if [ "$SKIP_REPO_UPDATE" = false ]; then
         exit 1
     fi
 
-    echo "📥 正在执行 git pull..."
-    if git pull; then
+    before_revision="$(git rev-parse HEAD)"
+    echo "📥 正在执行 git pull --ff-only..."
+    if git pull --ff-only; then
         echo "✅ 仓库已更新到最新版本"
     else
         echo "❌ git pull 失败，请检查网络或远程仓库配置"
@@ -116,6 +121,12 @@ if [ "$SKIP_REPO_UPDATE" = false ]; then
     [ ! -f .env.ASG ] && [ -f .env.ASG.sample ] && cp .env.ASG.sample .env.ASG && echo "   ✓ .env.ASG"
     [ ! -f .env.AMS ] && [ -f .env.AMS.sample ] && cp .env.AMS.sample .env.AMS && echo "   ✓ .env.AMS"
 
+    after_revision="$(git rev-parse HEAD)"
+    if [ "$before_revision" != "$after_revision" ] && [ "${EASYAI_UPDATE_REEXECUTED:-0}" != "1" ]; then
+        echo "🔄 检测到部署仓库已更新，切换到新版更新脚本继续执行..."
+        exec env EASYAI_UPDATE_REEXECUTED=1 UPDATE_SKIP_REPO_UPDATE=true "$SCRIPT_DIR/update.sh"
+    fi
+
     echo ""
 else
     echo "⏭️  跳过仓库更新，仅更新镜像并重启"
@@ -129,6 +140,10 @@ if [ ! -f .env ]; then
     exit 1
 fi
 
+# 每次更新都执行版本化、幂等的环境配置迁移。迁移只补缺失/空值或必要的列表项，
+# 不覆盖用户已经配置的非空值。后续新增必需配置时只需增加编号迁移文件。
+bash ./scripts/apply-update-env-migrations.sh .env "$SCRIPT_DIR"
+
 # 老部署缺少绑定配置时，根据现有公开 API 地址迁移：域名/本机保持回环，IP 直连保持可用。
 # shellcheck source=scripts/init-server-http-bind-ip.sh
 . ./scripts/init-server-http-bind-ip.sh
@@ -162,7 +177,7 @@ else
     sudo apt install -y apt-transport-https ca-certificates curl software-properties-common gnupg lsb-release
 
     echo "🔑 添加 Docker GPG 密钥..."
-    curl -fsSL https://mirrors.nwafu.edu.cn生成注释ker-ce/linux/ubuntu/gpg | sudo apt-key add -
+    curl -fsSL https://mirrors.nwafu.edu.cn/docker-ce/linux/ubuntu/gpg | sudo apt-key add -
 
     echo "🌍 添加 Docker 源..."
     sudo add-apt-repository -y "deb [arch=amd64] https://mirrors.nwafu.edu.cn生成注释ker-ce/linux/ubuntu/ $(lsb_release -cs) stable"
@@ -203,9 +218,65 @@ $DOCKER_COMPOSE_CMD version
 
 echo "🎉 Docker 和 Docker Compose 已就绪！"
 
+wait_for_core_services() {
+    local timeout_seconds="${UPDATE_HEALTH_TIMEOUT_SECONDS:-300}"
+    local deadline=$((SECONDS + timeout_seconds))
+    local services=(easyai-server ws-gateway easyai-web)
+    local service container_id state running health all_ready
+
+    if ! [[ "$timeout_seconds" =~ ^[1-9][0-9]*$ ]]; then
+        echo "❌ UPDATE_HEALTH_TIMEOUT_SECONDS 必须是正整数，当前为: $timeout_seconds"
+        return 1
+    fi
+
+    echo "⏳ 正在等待核心服务恢复可用（最长 ${timeout_seconds} 秒）..."
+    while (( SECONDS < deadline )); do
+        all_ready=true
+        for service in "${services[@]}"; do
+            container_id="$(sudo $DOCKER_COMPOSE_CMD ps -q "$service" 2>/dev/null || true)"
+            if [ -z "$container_id" ]; then
+                all_ready=false
+                continue
+            fi
+
+            state="$(sudo docker inspect --format '{{.State.Running}}|{{if .State.Health}}{{.State.Health.Status}}{{else}}none{{end}}' "$container_id" 2>/dev/null || true)"
+            running="${state%%|*}"
+            health="${state#*|}"
+            if [ "$running" != "true" ] || { [ "$health" != "none" ] && [ "$health" != "healthy" ]; }; then
+                all_ready=false
+            fi
+        done
+
+        if [ "$all_ready" = true ]; then
+            echo "✅ 核心服务已恢复可用"
+            return 0
+        fi
+        sleep 5
+    done
+
+    echo "❌ 更新后核心服务未在 ${timeout_seconds} 秒内恢复可用"
+    echo "   当前状态："
+    sudo $DOCKER_COMPOSE_CMD ps easyai-server ws-gateway easyai-web || true
+    echo "   请运行 '$DOCKER_COMPOSE_CMD logs --tail=100 easyai-server ws-gateway easyai-web' 查看原因"
+    return 1
+}
+
 # 旧 Redis 若仍把 AOF/RDB 留在容器写层，直接 up 会丢失数据。
 sudo ./scripts/check-redis-persistence.sh
 
 echo "🚀 重新启动EasyAI"
-sudo $DOCKER_COMPOSE_CMD pull && sudo $DOCKER_COMPOSE_CMD up -d
-echo "🎉EasyAI应用更新成功"
+echo "📦 当前磁盘空间："
+df -h "$SCRIPT_DIR" | tail -n 1
+if ! sudo $DOCKER_COMPOSE_CMD pull; then
+    echo "❌ 镜像拉取失败，EasyAI 未被标记为更新成功"
+    echo "   请检查网络、镜像仓库权限和服务器磁盘空间后重试"
+    exit 1
+fi
+if ! sudo $DOCKER_COMPOSE_CMD up -d; then
+    echo "❌ 容器启动失败，EasyAI 未被标记为更新成功"
+    exit 1
+fi
+if ! wait_for_core_services; then
+    exit 1
+fi
+echo "🎉 EasyAI 应用更新成功"