CI/CD 안정성 강화: 동시 배포 방지, 자기 재빌드 제거, 헬스체크 추가

- deploy.sh: flock으로 동시 배포 방지, deployer를 빌드 대상에서 제외
- deploy.sh: 배포 후 헬스체크 (4개 서비스 /health 확인)
- deploy.sh: 릴리즈 백업 최근 5개만 유지, 원자적 백업 (mv)
- deploy-nas.sh: .env 동기화 제거 (운영 시크릿 보호), __pycache__ 제외
- deployer: threading.Lock으로 동시 배포 방어, TimeoutExpired 개별 처리
- docker-compose: deployer 포트 localhost 바인딩, stock-lab 환경변수 추가

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

.env.example

@@ -50,3 +50,12 @@ PGID=1000
 
 # Windows AI Server (NAS 입장에서 바라본 Windows PC IP)
 WINDOWS_AI_SERVER_URL=http://192.168.45.59:8000
+
+# Admin API Key (trade/order 등 민감 엔드포인트 보호, 미설정 시 인증 비활성화)
+ADMIN_API_KEY=
+
+# Anthropic API Key (AI Coach 프록시, 미설정 시 AI Coach 비활성화)
+ANTHROPIC_API_KEY=
+
+# CORS 허용 도메인 (콤마 구분)
+CORS_ALLOW_ORIGINS=https://gahusb.synology.me,http://localhost:3007,http://localhost:8080

backend/.dockerignore

@@ -0,0 +1,6 @@
+.git
+__pycache__
+*.pyc
+.env
+.env.*
+*.md

deployer/.dockerignore

@@ -0,0 +1,6 @@
+.git
+__pycache__
+*.pyc
+.env
+.env.*
+*.md

deployer/app.py

@@ -1,14 +1,18 @@
-import os, hmac, hashlib, subprocess
+import os, hmac, hashlib, subprocess, threading
 from fastapi import FastAPI, Request, HTTPException, BackgroundTasks
 import logging
 
-# 로깅 설정
+logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(name)s] %(levelname)s %(message)s")
-logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger("deployer")
 
 app = FastAPI()
 SECRET = os.getenv("WEBHOOK_SECRET", "")
 
+if not SECRET:
+    logger.warning("WEBHOOK_SECRET is not set! All webhooks will be rejected.")
+
+_deploy_lock = threading.Lock()
+
 def verify(sig: str, body: bytes) -> bool:
     if not SECRET or not sig:
         return False
@@ -18,10 +22,13 @@ def verify(sig: str, body: bytes) -> bool:
     return any(hmac.compare_digest(sig, c) for c in candidates)
 
 def run_deploy_script():
-    """배포 스크립트를 백그라운드에서 실행하고 로그를 남김"""
+    """배포 스크립트를 백그라운드에서 실행 (동시 실행 방지)"""
-    logger.info("Starting deployment script...")
+    if not _deploy_lock.acquire(blocking=False):
+        logger.info("Deploy already in progress, skipping")
+        return
+
     try:
-        # 타임아웃 10분 설정
+        logger.info("Starting deployment script...")
         p = subprocess.run(["/bin/bash", "/scripts/deploy.sh"], capture_output=True, text=True, timeout=600)
 
         if p.returncode == 0:
@@ -29,8 +36,16 @@ def run_deploy_script():
         else:
             logger.error(f"Deployment FAILED ({p.returncode}):\n{p.stdout}\n{p.stderr}")
 
+    except subprocess.TimeoutExpired:
+        logger.error("Deployment TIMEOUT (10 min exceeded)")
     except Exception as e:
         logger.exception(f"Exception during deployment: {e}")
+    finally:
+        _deploy_lock.release()
+
+@app.get("/health")
+def health():
+    return {"status": "healthy", "service": "deployer"}
 
 @app.post("/webhook")
 async def webhook(req: Request, background_tasks: BackgroundTasks):

docker-compose.yml

@@ -31,6 +31,9 @@ services:
       - WINDOWS_AI_SERVER_URL=${WINDOWS_AI_SERVER_URL:-http://192.168.0.5:8000}
       - GEMINI_API_KEY=${GEMINI_API_KEY:-}
       - GEMINI_MODEL=${GEMINI_MODEL:-gemini-1.5-flash}
+      - ADMIN_API_KEY=${ADMIN_API_KEY:-}
+      - ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY:-}
+      - CORS_ALLOW_ORIGINS=${CORS_ALLOW_ORIGINS:-http://localhost:3007,http://localhost:8080}
     volumes:
       - ${STOCK_DATA_PATH:-./data/stock}:/app/data
 
@@ -45,6 +48,7 @@ services:
       - TZ=${TZ:-Asia/Seoul}
       - MUSIC_AI_SERVER_URL=${MUSIC_AI_SERVER_URL:-}
       - MUSIC_MEDIA_BASE=${MUSIC_MEDIA_BASE:-/media/music}
+      - CORS_ALLOW_ORIGINS=${CORS_ALLOW_ORIGINS:-http://localhost:3007,http://localhost:8080}
     volumes:
       - ${MUSIC_DATA_PATH:-./data/music}:/app/data
 
@@ -88,7 +92,7 @@ services:
     container_name: webpage-deployer
     restart: unless-stopped
     ports:
-      - "19010:9000"   # 외부 노출 필요 없으면 내부만 (리버스프록시로 /webhook만 노출 추천)
+      - "127.0.0.1:19010:9000"   # localhost만 허용 (nginx /webhook 프록시 경유)
     environment:
       - WEBHOOK_SECRET=${WEBHOOK_SECRET}
     volumes:

music-lab/.dockerignore

@@ -0,0 +1,6 @@
+.git
+__pycache__
+*.pyc
+.env
+.env.*
+*.md

music-lab/app/main.py

@@ -16,11 +16,13 @@ from .db import (
 
 app = FastAPI()
 
+_cors_origins = os.getenv("CORS_ALLOW_ORIGINS", "http://localhost:3007,http://localhost:8080").split(",")
 app.add_middleware(
     CORSMiddleware,
-    allow_origins=["*"],
+    allow_origins=[o.strip() for o in _cors_origins],
-    allow_methods=["*"],
+    allow_credentials=False,
-    allow_headers=["*"],
+    allow_methods=["GET", "POST", "PUT", "DELETE", "OPTIONS"],
+    allow_headers=["Content-Type"],
 )
 
 MUSIC_AI_SERVER_URL = os.getenv("MUSIC_AI_SERVER_URL", "")

scripts/deploy-nas.sh

@@ -30,33 +30,14 @@ cd "$SRC"
 
 # 레포에서 운영으로 반영할 항목들만 복사/동기화 (필요한 것만 적기)
 # backend, travel-proxy, deployer, nginx, scripts, docker-compose.yml, .env 등
-rsync -a --delete \
+RSYNC_EXCLUDES="--exclude .git --exclude __pycache__ --exclude *.pyc --exclude data/"
-  --exclude ".git" \
-  --exclude ".releases" \
-  "$SRC/backend/"        "$DST/backend/"
-rsync -a --delete \
-  --exclude ".git" \
-  "$SRC/travel-proxy/"   "$DST/travel-proxy/"
-rsync -a --delete \
-  --exclude ".git" \
-  "$SRC/deployer/"       "$DST/deployer/"
-rsync -a --delete \
-    --exclude ".git" \
-    "$SRC/stock-lab/"      "$DST/stock-lab/"
-rsync -a --delete \
-    --exclude ".git" \
-    "$SRC/music-lab/"      "$DST/music-lab/"
-rsync -a --delete \
-  --exclude ".git" \
-  "$SRC/nginx/"          "$DST/nginx/"
-rsync -a --delete \
-  --exclude ".git" \
-  "$SRC/scripts/"        "$DST/scripts/"
 
-# compose 파일 / env 파일
+for dir in backend travel-proxy deployer stock-lab music-lab nginx scripts; do
+    rsync -a --delete $RSYNC_EXCLUDES \
+      "$SRC/$dir/" "$DST/$dir/"
+done
+
+# compose 파일만 동기화 (.env는 절대 동기화하지 않음 — 운영 시크릿 보호)
 rsync -a "$SRC/docker-compose.yml" "$DST/docker-compose.yml"
-if [ -f "$SRC/.env" ]; then
-  rsync -a "$SRC/.env" "$DST/.env"
-fi
 
 echo "SYNC_OK"

scripts/deploy.sh

@@ -1,6 +1,10 @@
 #!/bin/bash
 set -euo pipefail
 
+# ── 동시 배포 방지 (flock) ──
+exec 200>/tmp/deploy.lock
+flock -n 200 || { echo "Deploy already running, skipping"; exit 0; }
+
 # 1. 자동 감지: Docker 컨테이너 내부인가?
 if [ -d "/repo" ] && [ -d "/runtime" ]; then
     echo "Detected Docker Container environment."
@@ -13,9 +17,8 @@ else
         set -a; source .env; set +a
     fi
 
-    # 환경변수가 없으면 현재 디렉토리를 SRC로
     SRC="${REPO_PATH:-$(pwd)}"
-    DST="${RUNTIME_PATH:-/volume1/docker/webpage}" # 기본값 설정
+    DST="${RUNTIME_PATH:-/volume1/docker/webpage}"
 
     if [ -z "$DST" ]; then
         echo "Error: RUNTIME_PATH is not set."
@@ -32,19 +35,40 @@ cd "$SRC"
 git fetch --all --prune
 git pull --ff-only
 
-# 릴리즈 백업(롤백용): 아래 5번과 연결
+# ── 릴리즈 백업 (롤백용) ──
 TAG="$(date +%Y%m%d-%H%M%S)"
-mkdir -p "$DST/.releases/$TAG"
+BACKUP_DIR="$DST/.releases/$TAG"
+mkdir -p "$BACKUP_DIR.tmp"
 rsync -a --delete \
   --exclude ".releases" \
-  "$DST/" "$DST/.releases/$TAG/"
+  "$DST/" "$BACKUP_DIR.tmp/"
+mv "$BACKUP_DIR.tmp" "$BACKUP_DIR"
 
-# 소스 → 운영 반영 (네가 이미 만든 deploy-nas.sh가 있으면 그걸 호출해도 됨)
+# 오래된 릴리즈 정리 (최근 5개만 유지)
-# 예: repo/scripts/deploy-nas.sh가 운영으로 복사/동기화하는 로직이라면:
+ls -dt "$DST/.releases"/*/ 2>/dev/null | tail -n +6 | xargs -r rm -rf
+
+# ── 소스 → 운영 반영 ──
 bash "$SRC/scripts/deploy-nas.sh"
 
+# ── 컨테이너 재빌드 (deployer 제외 — 자기 자신을 재빌드하면 스크립트 중단됨) ──
 cd "$DST"
-docker compose up -d --build backend travel-proxy stock-lab frontend deployer
+docker compose up -d --build backend travel-proxy stock-lab music-lab frontend
 docker exec lotto-frontend nginx -s reload 2>/dev/null || true
 
+# ── 배포 후 헬스체크 ──
+echo "Waiting for services to start..."
+sleep 5
+
+HEALTH_OK=true
+for endpoint in "http://backend:8000/health" "http://stock-lab:8000/health" "http://travel-proxy:8000/health" "http://music-lab:8000/health"; do
+    if ! curl -sf --max-time 5 "$endpoint" > /dev/null 2>&1; then
+        echo "HEALTH_FAIL: $endpoint"
+        HEALTH_OK=false
+    fi
+done
+
+if [ "$HEALTH_OK" = false ]; then
+    echo "DEPLOY_WARN: Some services failed health check. Consider rollback: $TAG"
+else
     echo "DEPLOY_OK $TAG"
+fi

stock-lab/.dockerignore

@@ -0,0 +1,6 @@
+.git
+__pycache__
+*.pyc
+.env
+.env.*
+*.md

stock-lab/app/main.py

@@ -1,14 +1,18 @@
 import os
 import json
+import logging
 from datetime import date as date_type
 from typing import Optional
-from fastapi import FastAPI, Query
+from fastapi import FastAPI, Query, Header, Depends, HTTPException
 from fastapi.responses import JSONResponse
 from fastapi.middleware.cors import CORSMiddleware
 import requests
 from apscheduler.schedulers.background import BackgroundScheduler
 from pydantic import BaseModel
 
+logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(name)s] %(levelname)s %(message)s")
+logger = logging.getLogger("stock-lab")
+
 from .db import (
     init_db, save_articles, get_latest_articles,
     add_portfolio_item, get_all_portfolio, get_portfolio_item,
@@ -23,12 +27,13 @@ from .price_fetcher import get_current_prices
 app = FastAPI()
 
 # CORS 설정 (프론트엔드 접근 허용)
+_cors_origins = os.getenv("CORS_ALLOW_ORIGINS", "http://localhost:3007,http://localhost:8080").split(",")
 app.add_middleware(
     CORSMiddleware,
-    allow_origins=["*"],  # 운영 시에는 구체적인 도메인으로 제한하는 것이 좋음
+    allow_origins=[o.strip() for o in _cors_origins],
-    allow_credentials=True,
+    allow_credentials=False,
-    allow_methods=["*"],
+    allow_methods=["GET", "POST", "PUT", "DELETE", "OPTIONS"],
-    allow_headers=["*"],
+    allow_headers=["Content-Type"],
 )
 
 scheduler = BackgroundScheduler(timezone=os.getenv("TZ", "Asia/Seoul"))
@@ -36,6 +41,19 @@ scheduler = BackgroundScheduler(timezone=os.getenv("TZ", "Asia/Seoul"))
 # Windows AI Server URL (NAS .env에서 설정)
 WINDOWS_AI_SERVER_URL = os.getenv("WINDOWS_AI_SERVER_URL", "http://192.168.0.5:8000")
 
+# Admin API Key 인증
+ADMIN_API_KEY = os.getenv("ADMIN_API_KEY", "")
+
+def verify_admin(x_admin_key: str = Header(None)):
+    """admin/trade 엔드포인트 보호용 API 키 검증"""
+    if not ADMIN_API_KEY:
+        return  # 키 미설정 시 인증 비활성화 (개발 환경)
+    if x_admin_key != ADMIN_API_KEY:
+        raise HTTPException(status_code=401, detail="Unauthorized")
+
+# Anthropic API 프록시용 키 (서버 측 보관)
+ANTHROPIC_API_KEY = os.getenv("ANTHROPIC_API_KEY", "")
+
 # 공휴일 목록 로드
 _HOLIDAYS_PATH = os.path.join(os.path.dirname(__file__), "holidays.json")
 try:
@@ -123,53 +141,87 @@ def trigger_scrap():
     run_scraping_job()
     return {"ok": True}
 
-# --- Trading API (Windows Proxy) ---
+# --- Trading API (Windows Proxy, 인증 필요) ---
 
-@app.get("/api/trade/balance")
+@app.get("/api/trade/balance", dependencies=[Depends(verify_admin)])
 def get_balance():
     """계좌 잔고 조회 (Windows AI Server Proxy)"""
-    print(f"[Proxy] Requesting Balance from {WINDOWS_AI_SERVER_URL}...")
+    logger.info(f"Requesting Balance from {WINDOWS_AI_SERVER_URL}")
     try:
         resp = requests.get(f"{WINDOWS_AI_SERVER_URL}/trade/balance", timeout=5)
-        
         if resp.status_code != 200:
-            print(f"[ProxyError] Balance Error: {resp.status_code} {resp.text}")
+            logger.error(f"Balance Error: {resp.status_code}")
             return JSONResponse(status_code=resp.status_code, content=resp.json())
-            
-        print("[Proxy] Balance Success")
         return resp.json()
+    except requests.JSONDecodeError:
+        return JSONResponse(status_code=resp.status_code, content={"error": f"Upstream error {resp.status_code}"})
     except Exception as e:
-        print(f"[ProxyError] Connection Failed: {e}")
+        logger.error(f"Balance Connection Failed: {e}")
-        return JSONResponse(
+        return JSONResponse(status_code=500, content={"error": "Connection Failed"})
-            status_code=500, 
-            content={"error": "Connection Failed", "detail": str(e), "target": WINDOWS_AI_SERVER_URL}
-        )
 
 class OrderRequest(BaseModel):
-    ticker: str         # 종목 코드 (예: "005930")
+    ticker: str
-    action: str         # "BUY" or "SELL"
+    action: str
-    quantity: int       # 주문 수량
+    quantity: int
-    price: int = 0      # 0이면 시장가
+    price: int = 0
-    reason: Optional[str] = "Manual Order" # 주문 사유 (AI 기록용)
+    reason: Optional[str] = "Manual Order"
 
-@app.post("/api/trade/order")
+@app.post("/api/trade/order", dependencies=[Depends(verify_admin)])
 def order_stock(req: OrderRequest):
     """주식 매수/매도 주문 (Windows AI Server Proxy)"""
-    print(f"[Proxy] Order Request: {req.dict()} to {WINDOWS_AI_SERVER_URL}...")
+    logger.info(f"Order Request: {req.action} {req.ticker} x{req.quantity}")
     try:
         resp = requests.post(f"{WINDOWS_AI_SERVER_URL}/trade/order", json=req.dict(), timeout=10)
-        
         if resp.status_code != 200:
-            print(f"[ProxyError] Order Error: {resp.status_code} {resp.text}")
+            logger.error(f"Order Error: {resp.status_code}")
             return JSONResponse(status_code=resp.status_code, content=resp.json())
-            
         return resp.json()
+    except requests.JSONDecodeError:
+        return JSONResponse(status_code=resp.status_code, content={"error": f"Upstream error {resp.status_code}"})
     except Exception as e:
-        print(f"[ProxyError] Order Connection Failed: {e}")
+        logger.error(f"Order Connection Failed: {e}")
-        return JSONResponse(
+        return JSONResponse(status_code=500, content={"error": "Connection Failed"})
-            status_code=500,
+
-            content={"error": "Connection Failed", "detail": str(e), "target": WINDOWS_AI_SERVER_URL}
+# --- AI Coach 프록시 (API 키를 서버에 보관) ---
+
+class AiCoachRequest(BaseModel):
+    model: str = "claude-haiku-4-5-20251001"
+    prompt: str
+    max_tokens: int = 1024
+
+@app.post("/api/stock/ai-coach")
+def ai_coach(req: AiCoachRequest):
+    """AI 포트폴리오 코치 — Anthropic API 프록시 (API 키 서버 보관)"""
+    if not ANTHROPIC_API_KEY:
+        raise HTTPException(503, "AI Coach not configured (ANTHROPIC_API_KEY missing)")
+
+    allowed_models = {"claude-haiku-4-5-20251001", "claude-sonnet-4-6"}
+    model = req.model if req.model in allowed_models else "claude-haiku-4-5-20251001"
+
+    try:
+        resp = requests.post(
+            "https://api.anthropic.com/v1/messages",
+            headers={
+                "Content-Type": "application/json",
+                "x-api-key": ANTHROPIC_API_KEY,
+                "anthropic-version": "2023-06-01",
+            },
+            json={
+                "model": model,
+                "max_tokens": req.max_tokens,
+                "messages": [{"role": "user", "content": req.prompt}],
+            },
+            timeout=30,
         )
+        if resp.status_code != 200:
+            logger.error(f"Anthropic API error: {resp.status_code}")
+            return JSONResponse(status_code=resp.status_code, content={"error": "AI API error"})
+        return resp.json()
+    except requests.Timeout:
+        return JSONResponse(status_code=504, content={"error": "AI API timeout"})
+    except Exception as e:
+        logger.error(f"AI Coach error: {e}")
+        return JSONResponse(status_code=500, content={"error": "AI Coach failed"})
 
 
 

travel-proxy/.dockerignore

@@ -0,0 +1,6 @@
+.git
+__pycache__
+*.pyc
+.env
+.env.*
+*.md

travel-proxy/app/main.py

@@ -168,6 +168,10 @@ def scan_album(album: str) -> List[Dict[str, Any]]:
 # -----------------------------
 # Routes
 # -----------------------------
+@app.get("/health")
+def health():
+    return {"status": "healthy", "service": "travel-proxy"}
+
 @app.get("/api/travel/regions")
 def regions():
     _meta_changed_invalidate_cache()
@@ -239,12 +243,18 @@ def photos(
 
 @app.get("/media/travel/.thumb/{album}/{filename}")
 def get_thumb(album: str, filename: str):
+    if ".." in album or ".." in filename:
+        raise HTTPException(400, "Invalid path")
+
     src = (ROOT / album / filename).resolve()
+    if not str(src).startswith(str(ROOT)):
+        raise HTTPException(403, "Access denied")
+    if not src.exists() or not src.is_file():
+        raise HTTPException(404, "Source not found")
+
+    p = ensure_thumb(src, album)
     if not p.exists() or not p.is_file():
         raise HTTPException(404, "Thumbnail not found")
-
-    # src로부터 thumb 생성/확인 (원본 확장자 유지)
-    p = ensure_thumb(src, album)
     return FileResponse(str(p))
 
 @app.get("/api/version")