diff --git a/.env.server.example b/.env.server.example
index cd81f424..3f207db0 100644
--- a/.env.server.example
+++ b/.env.server.example
@@ -40,16 +40,18 @@ LOG_LEVEL=info
 LOG_INTEGRATION_LEVEL=warn
 LOG_RETENTION_DAYS=30
 
-# 文件存储：服务器迁移期间设 dual；校验通过、停止 MinIO 前改为 oss。
-STORAGE_MODE=minio
+# 文件存储：生产使用阿里云 OSS（生产编排已移除 MinIO，仅本地开发使用）。
+# STORAGE_MODE 取值 minio | dual | oss；dual 仅在 MinIO→OSS 迁移/回滚期间使用。
+STORAGE_MODE=oss
+# 以下 MinIO 配置仅在 STORAGE_MODE=dual/minio（迁移回滚场景）时使用。
 STORAGE_ENDPOINT=http://minio:9000
 STORAGE_BUCKET=order-site
 STORAGE_ACCESS_KEY_ID=change-me-storage-access-key
 STORAGE_SECRET_ACCESS_KEY=change-me-storage-secret-key
 STORAGE_REGION=us-east-1
 STORAGE_MAX_UPLOAD_SIZE_MB=10
-# 阿里云 OSS S3 兼容端点（不要填写 bucket 前缀）
-STORAGE_OSS_ENDPOINT=https://oss-cn-hangzhou.aliyuncs.com
+# 阿里云 OSS S3 兼容端点：ECS 与桶同地域时用内网地址（免流量费、低延迟）
+STORAGE_OSS_ENDPOINT=https://oss-cn-hangzhou-internal.aliyuncs.com
 STORAGE_OSS_BUCKET=change-me-oss-bucket
 STORAGE_OSS_ACCESS_KEY_ID=change-me-oss-access-key
 STORAGE_OSS_SECRET_ACCESS_KEY=change-me-oss-secret-key
diff --git a/deploy/ubuntu-deploy.sh b/deploy/ubuntu-deploy.sh
index b950a4af..1e19acf9 100755
--- a/deploy/ubuntu-deploy.sh
+++ b/deploy/ubuntu-deploy.sh
@@ -16,11 +16,6 @@ HEALTH_URL=""
 READY_TIMEOUT="${READY_TIMEOUT:-120}"
 RESET_DB=false
 RESET_DB_CONFIRM="${RESET_DB_CONFIRM:-}"
-# 激活 Compose profile:默认启用 storage,让内置 MinIO 随部署自动启动。
-# 若改用外部 S3/OSS,可在环境变量前置 COMPOSE_PROFILES= 赋空值跳过内置 MinIO。
-if [[ -z "${COMPOSE_PROFILES+x}" ]]; then
-  COMPOSE_PROFILES=(--profile storage)
-fi
 
 log() {
   printf '\033[1;34m[deploy]\033[0m %s\n' "$*"
@@ -51,8 +46,7 @@ print_usage() {
   日常结构变更请新增 002_xxx.sql 等渐进迁移，不要改已应用的历史迁移文件。
   它不会删除 apps/backend/data 下的平台 JSON 配置，也不会删除 Caddy 证书卷。
 
-  默认激活 storage profile 启动内置 MinIO。若使用外部 S3/OSS,可在环境变量前置
-  COMPOSE_PROFILES= 赋空值跳过内置 MinIO,例如:COMPOSE_PROFILES= bash deploy/ubuntu-deploy.sh
+  图片存储使用阿里云 OSS（.env 的 STORAGE_OSS_*），生产编排不含 MinIO。
 EOF
 }
 
@@ -95,11 +89,6 @@ ensure_docker() {
   command -v docker >/dev/null 2>&1 || fail "未找到 docker，请先在 Ubuntu 上安装 Docker Engine。"
   docker compose version >/dev/null 2>&1 || fail "未找到 docker compose 插件，请安装 Docker Compose plugin。"
   command -v curl >/dev/null 2>&1 || fail "未找到 curl，请先安装：sudo apt-get update && sudo apt-get install -y curl"
-
-  # 允许通过环境变量 COMPOSE_PROFILES= 跳过内置 MinIO(用外部 S3/OSS 时)。
-  if [[ "${#COMPOSE_PROFILES[@]}" -eq 0 ]]; then
-    log "COMPOSE_PROFILES 为空,跳过内置 MinIO(请确保 .env 的 STORAGE_ENDPOINT 指向外部存储)"
-  fi
 }
 
 ensure_env_file() {
@@ -197,20 +186,20 @@ ensure_data_dir() {
 
 check_compose_config() {
   log "检查 Docker Compose 配置"
-  docker compose "${COMPOSE_PROFILES[@]}" config >/dev/null || fail "Docker Compose 配置校验失败。"
+  docker compose config >/dev/null || fail "Docker Compose 配置校验失败。"
 }
 
 pull_base_images() {
   log "拉取基础镜像"
-  docker compose "${COMPOSE_PROFILES[@]}" pull --ignore-buildable
+  docker compose pull --ignore-buildable
 }
 
 build_images() {
   log "构建后端镜像"
-  docker compose "${COMPOSE_PROFILES[@]}" build backend || fail "后端镜像构建失败。"
+  docker compose build backend || fail "后端镜像构建失败。"
 
   log "构建 Web 镜像"
-  docker compose "${COMPOSE_PROFILES[@]}" build web || fail "Web 镜像构建失败。"
+  docker compose build web || fail "Web 镜像构建失败。"
 }
 
 resolve_compose_project_name() {
@@ -227,7 +216,7 @@ resolve_compose_project_name() {
 resolve_postgres_volume_name() {
   local container_id volume_name project_name
 
-  container_id="$(docker compose "${COMPOSE_PROFILES[@]}" ps -aq postgres 2>/dev/null | head -n 1 || true)"
+  container_id="$(docker compose ps -aq postgres 2>/dev/null | head -n 1 || true)"
   if [[ -n "${container_id}" ]]; then
     volume_name="$(
       docker inspect -f '{{range .Mounts}}{{if eq .Destination "/var/lib/postgresql/data"}}{{if eq .Type "volume"}}{{.Name}}{{end}}{{end}}{{end}}' "${container_id}" 2>/dev/null || true
@@ -239,7 +228,7 @@ resolve_postgres_volume_name() {
   fi
 
   volume_name="$(
-    docker compose "${COMPOSE_PROFILES[@]}" volumes 2>/dev/null \
+    docker compose volumes 2>/dev/null \
       | awk '$2 ~ /(^|_)postgres_data$/ { print $2; exit }' || true
   )"
   if [[ -n "${volume_name}" ]]; then
@@ -260,8 +249,8 @@ reset_database() {
   volume_name="$(resolve_postgres_volume_name)"
 
   warn "开始重置数据库，将删除 Docker volume：${volume_name}"
-  docker compose "${COMPOSE_PROFILES[@]}" stop web backend postgres >/dev/null 2>&1 || true
-  docker compose "${COMPOSE_PROFILES[@]}" rm -f -s postgres >/dev/null 2>&1 || true
+  docker compose stop web backend postgres >/dev/null 2>&1 || true
+  docker compose rm -f -s postgres >/dev/null 2>&1 || true
 
   if docker volume inspect "${volume_name}" >/dev/null 2>&1; then
     docker volume rm "${volume_name}" >/dev/null || fail "删除 PostgreSQL 数据卷失败：${volume_name}"
@@ -280,7 +269,7 @@ wait_for_service() {
   log "等待 ${label} 就绪"
 
   for ((attempt = 1; attempt <= READY_TIMEOUT; attempt++)); do
-    container_id="$(docker compose "${COMPOSE_PROFILES[@]}" ps -q "${service}" 2>/dev/null || true)"
+    container_id="$(docker compose ps -q "${service}" 2>/dev/null || true)"
     if [[ -n "${container_id}" ]]; then
       status="$(docker inspect -f '{{.State.Status}}' "${container_id}" 2>/dev/null || true)"
       health_status="$(docker inspect -f '{{if .State.Health}}{{.State.Health.Status}}{{end}}' "${container_id}" 2>/dev/null || true)"
@@ -309,7 +298,7 @@ wait_for_service() {
 
 start_postgres() {
   log "启动 PostgreSQL"
-  if docker compose "${COMPOSE_PROFILES[@]}" up -d --no-deps postgres; then
+  if docker compose up -d --no-deps postgres; then
     wait_for_service postgres "PostgreSQL"
     return
   fi
@@ -318,30 +307,9 @@ start_postgres() {
   fail "PostgreSQL 启动失败，请根据上面的日志排查。"
 }
 
-start_minio() {
-  # 未激活 storage profile 时跳过内置 MinIO(用外部 S3/OSS 的场景)。
-  local has_storage_profile=false
-  for arg in "${COMPOSE_PROFILES[@]}"; do
-    [[ "${arg}" == "storage" ]] && has_storage_profile=true
-  done
-  if [[ "${has_storage_profile}" != "true" ]]; then
-    log "跳过内置 MinIO(COMPOSE_PROFILES 未包含 storage)"
-    return
-  fi
-
-  log "启动 MinIO"
-  if docker compose "${COMPOSE_PROFILES[@]}" up -d --no-deps minio; then
-    wait_for_service minio "MinIO"
-    return
-  fi
-
-  print_diagnostics
-  fail "MinIO 启动失败,请根据上面的日志排查。"
-}
-
 run_database_migrations() {
   log "执行数据库迁移"
-  if docker compose "${COMPOSE_PROFILES[@]}" run --rm --no-deps backend node dist/db/migrate.js; then
+  if docker compose run --rm --no-deps backend node dist/db/migrate.js; then
     log "数据库迁移完成"
     return
   fi
@@ -352,7 +320,7 @@ run_database_migrations() {
 
 restart_backend() {
   log "重建并启动后端"
-  if docker compose "${COMPOSE_PROFILES[@]}" up -d --no-deps --force-recreate backend; then
+  if docker compose up -d --no-deps --force-recreate backend; then
     wait_for_service backend "后端"
     return
   fi
@@ -363,7 +331,7 @@ restart_backend() {
 
 restart_web() {
   log "重建并启动 Web 入口"
-  if docker compose "${COMPOSE_PROFILES[@]}" up -d --no-deps --force-recreate --remove-orphans web; then
+  if docker compose up -d --no-deps --force-recreate --remove-orphans web; then
     wait_for_service web "Web"
     return
   fi
@@ -378,7 +346,6 @@ deploy_stack() {
   build_images
   reset_database
   start_postgres
-  start_minio
   run_database_migrations
   restart_backend
   restart_web
@@ -403,7 +370,7 @@ print_applied_migrations() {
   log "已应用数据库迁移"
 
   local migrations
-  if migrations="$(docker compose "${COMPOSE_PROFILES[@]}" exec -T backend node --input-type=module -e '
+  if migrations="$(docker compose exec -T backend node --input-type=module -e '
 const { query, closeDb } = await import("./dist/db/client.js");
 try {
   const result = await query("SELECT filename FROM schema_migrations ORDER BY filename");
@@ -422,11 +389,11 @@ try {
 
 print_diagnostics() {
   warn "当前容器状态："
-  docker compose "${COMPOSE_PROFILES[@]}" ps || true
+  docker compose ps || true
 
   warn "后端最近日志："
   local backend_logs
-  backend_logs="$(docker compose "${COMPOSE_PROFILES[@]}" logs --tail=120 backend 2>&1 || true)"
+  backend_logs="$(docker compose logs --tail=120 backend 2>&1 || true)"
   printf '%s\n' "${backend_logs}"
 
   if printf '%s\n' "${backend_logs}" | grep -q 'password authentication failed for user "postgres"'; then
@@ -436,7 +403,7 @@ print_diagnostics() {
   fi
 
   warn "后端 ready 接口："
-  docker compose "${COMPOSE_PROFILES[@]}" exec -T backend sh -lc 'curl -fsS http://127.0.0.1:3000/health/ready || true' || true
+  docker compose exec -T backend sh -lc 'curl -fsS http://127.0.0.1:3000/health/ready || true' || true
 
   print_applied_migrations || true
 }
@@ -456,7 +423,7 @@ main() {
   print_applied_migrations
 
   log "部署完成"
-  docker compose "${COMPOSE_PROFILES[@]}" ps
+  docker compose ps
 }
 
 main "$@"
diff --git a/docker-compose.yml b/docker-compose.yml
index 284cea60..21d4f820 100644
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -71,7 +71,8 @@ services:
       DATA_ROOT: /app/data
       LOG_RETENTION_DAYS: ${LOG_RETENTION_DAYS:-30}
       CLAIM_BASE_URL: ${CLAIM_BASE_URL:-http://localhost/#/claim}
-      STORAGE_MODE: ${STORAGE_MODE:-minio}
+      # 生产图片存储只走阿里云 OSS；MinIO 仅保留在本地开发编排（docker-compose.dev.yml）。
+      STORAGE_MODE: ${STORAGE_MODE:-oss}
       STORAGE_ENDPOINT: ${STORAGE_ENDPOINT:-http://minio:9000}
       STORAGE_BUCKET: ${STORAGE_BUCKET:-order-site}
       STORAGE_ACCESS_KEY_ID: ${STORAGE_ACCESS_KEY_ID:-minioadmin}
@@ -90,22 +91,6 @@ services:
     volumes:
       - ./apps/backend/data:/app/data
 
-  minio:
-    image: docker.m.daocloud.io/minio/minio:latest
-    profiles: ["storage"]
-    restart: unless-stopped
-    logging: *json-log-rotation
-    command: server /data --address ":9000" --console-address ":9001"
-    environment:
-      TZ: ${TZ:-Asia/Shanghai}
-      MINIO_ROOT_USER: ${STORAGE_ACCESS_KEY_ID:-minioadmin}
-      MINIO_ROOT_PASSWORD: ${STORAGE_SECRET_ACCESS_KEY:-minioadmin}
-    ports:
-      - "9000:9000"
-      - "9001:9001"
-    volumes:
-      - minio_data:/data
-
   web:
     build:
       context: .
@@ -134,6 +119,5 @@ services:
 
 volumes:
   postgres_data:
-  minio_data:
   caddy_data:
   caddy_config:
diff --git a/docs/服务器图片迁移到阿里云OSS.md b/docs/服务器图片迁移到阿里云OSS.md
index e5070278..97c47981 100644
--- a/docs/服务器图片迁移到阿里云OSS.md
+++ b/docs/服务器图片迁移到阿里云OSS.md
@@ -46,13 +46,26 @@ curl -fsS https://你的域名/health/ready
 
 此时新旧图片都只从 OSS 读取，数据库中的对象 key 和访问 URL 不需要改动。
 
-## 4. 停止 MinIO
+## 4. MinIO 退场
 
-先观察 OSS 模式运行一段时间，确认图片上传、后台预览、打手端验收图片均正常，再停止并移除 MinIO 容器。不要删除 `minio_data` 卷，至少保留一个回滚窗口：
+先观察 OSS 模式运行一段时间，确认图片上传、后台预览、打手端验收图片均正常，再停止并移除 MinIO 容器（若仍由旧版编排管理）：
 
 ```bash
 docker compose --profile storage stop minio
 docker compose --profile storage rm -f minio
 ```
 
-若需要回滚，恢复 `STORAGE_MODE=dual`（或 `minio`），重新启动 MinIO 即可。确认 OSS 已稳定且备份完成后，再单独评估删除 `minio_data` 卷。
+生产编排（`docker-compose.yml` 与 `deploy/ubuntu-deploy.sh`）已移除 MinIO 服务与 `storage` profile，此后部署不会再启动 MinIO。
+
+不要删除 `minio_data` 卷（实际名称 `order_site_minio_data`），至少保留一个回滚窗口。若需要回滚，把 `.env` 的 `STORAGE_MODE` 改回 `dual`，并手动用旧卷拉起一个 MinIO 容器接入 compose 网络（`--network-alias minio` 保证 backend 仍可用 `http://minio:9000` 访问它）：
+
+```bash
+docker run -d --name minio-rollback \
+  --network order_site_default \
+  --network-alias minio \
+  -v order_site_minio_data:/data \
+  docker.m.daocloud.io/minio/minio:latest \
+  server /data --address ":9000" --console-address ":9001"
+```
+
+确认 OSS 已稳定且完成密钥轮换后，再单独评估删除 `minio_data` 卷释放磁盘。