feat(packs-lab): DSM 호출 retry/backoff + 업로드 cleanup 보강

- dsm_client.py: _request_with_retry()로 5xx·transport·timeout만 지수백오프
  재시도 (DSM_MAX_RETRIES, DSM_BACKOFF_SEC env). DSM error code 응답 본문 로깅.
- routes.py: upload 핸들러를 try/finally로 감싸 부분파일 정리 보장, Supabase
  INSERT 호출 자체에 try/except 추가해 네트워크 예외도 cleanup.
- test_dsm_client.py: retry 시나리오 4종 추가 (5xx→성공/소진/transport
  error/4xx no-retry). 전체 29 테스트 pass.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

packs-lab/app/dsm_client.py

@@ -4,6 +4,7 @@
 
 - create_share_link(file_path, expires_in_sec) -> share URL
 """
+import asyncio
 import logging
 import os
 from datetime import datetime, timedelta, timezone
@@ -18,6 +19,8 @@ DSM_PASS = os.getenv("DSM_PASS", "")
 # LAN IP로 DSM 접근 시 self-signed cert가 IP에 매칭 안 되어 검증 실패. LAN 내부 통신이라 false 허용.
 # 운영에서 LAN IP + self-signed면 DSM_VERIFY_SSL=false. 도메인 + 정상 cert면 기본값(true) 유지.
 DSM_VERIFY_SSL = os.getenv("DSM_VERIFY_SSL", "true").strip().lower() != "false"
+DSM_MAX_RETRIES = max(1, int(os.getenv("DSM_MAX_RETRIES", "3")))
+DSM_BACKOFF_SEC = float(os.getenv("DSM_BACKOFF_SEC", "0.5"))
 
 API_AUTH = "/webapi/auth.cgi"
 API_SHARE = "/webapi/entry.cgi"
@@ -27,13 +30,45 @@ class DSMError(RuntimeError):
     pass
 
 
+async def _request_with_retry(
+    client: httpx.AsyncClient,
+    url: str,
+    params: dict,
+    timeout: float,
+) -> httpx.Response:
+    """5xx · transport · timeout만 지수백오프 retry. 4xx와 DSM success=false는 호출자가 판단."""
+    last_exc: Exception | None = None
+    for attempt in range(DSM_MAX_RETRIES):
+        try:
+            r = await client.get(url, params=params, timeout=timeout)
+            if r.status_code < 500:
+                return r
+            last_exc = httpx.HTTPStatusError(
+                f"HTTP {r.status_code}", request=r.request, response=r
+            )
+            logger.warning(
+                "DSM HTTP %s — attempt %s/%s body=%s",
+                r.status_code, attempt + 1, DSM_MAX_RETRIES, r.text[:200],
+            )
+        except (httpx.TransportError, httpx.TimeoutException) as e:
+            last_exc = e
+            logger.warning(
+                "DSM transport error: %s — attempt %s/%s",
+                e, attempt + 1, DSM_MAX_RETRIES,
+            )
+        if attempt < DSM_MAX_RETRIES - 1:
+            await asyncio.sleep(DSM_BACKOFF_SEC * (2 ** attempt))
+    raise DSMError(f"DSM 요청 실패 (재시도 {DSM_MAX_RETRIES}회): {last_exc}")
+
+
 async def _login(client: httpx.AsyncClient) -> str:
     """DSM 세션 sid 반환."""
     if not all([DSM_HOST, DSM_USER, DSM_PASS]):
         raise DSMError("DSM 환경변수 미설정")
-    r = await client.get(
+    r = await _request_with_retry(
+        client,
         f"{DSM_HOST}{API_AUTH}",
-        params={
+        {
             "api": "SYNO.API.Auth",
             "version": "7",
             "method": "login",
@@ -42,12 +77,14 @@ async def _login(client: httpx.AsyncClient) -> str:
             "session": "FileStation",
             "format": "sid",
         },
-        timeout=15.0,
+        15.0,
     )
     r.raise_for_status()
     data = r.json()
     if not data.get("success"):
-        raise DSMError(f"DSM login 실패: {data.get('error')}")
+        err = data.get("error", {})
+        logger.error("DSM login 실패: code=%s error=%s", err.get("code"), err)
+        raise DSMError(f"DSM login 실패: code={err.get('code')} error={err}")
     return data["data"]["sid"]
 
 
@@ -80,9 +117,10 @@ async def create_share_link(file_path: str, expires_in_sec: int = 14400) -> tupl
     async with httpx.AsyncClient(verify=DSM_VERIFY_SSL) as client:
         sid = await _login(client)
         try:
-            r = await client.get(
+            r = await _request_with_retry(
+                client,
                 f"{DSM_HOST}{API_SHARE}",
-                params={
+                {
                     "api": "SYNO.FileStation.Sharing",
                     "version": "3",
                     "method": "create",
@@ -90,16 +128,22 @@ async def create_share_link(file_path: str, expires_in_sec: int = 14400) -> tupl
                     "date_expired": expire_time_ms,
                     "_sid": sid,
                 },
-                timeout=15.0,
+                15.0,
             )
             r.raise_for_status()
             data = r.json()
             if not data.get("success"):
-                raise DSMError(f"DSM Sharing.create 실패: {data.get('error')}")
+                err = data.get("error", {})
+                logger.error(
+                    "DSM Sharing.create 실패: path=%s code=%s error=%s",
+                    file_path, err.get("code"), err,
+                )
+                raise DSMError(f"DSM Sharing.create 실패: code={err.get('code')} error={err}")
             links = data["data"]["links"]
             if not links:
                 raise DSMError("Sharing 응답에 링크 없음")
             url = links[0]["url"]
+            logger.info("DSM share link created: path=%s", file_path)
             return url, expires_at
         finally:
             await _logout(client, sid)