fix: address code review findings

- Add retry (2 attempts) on CivitAI 429 rate limit instead of silently skipping
- Use check_same_thread=False for SQLite connection (aiohttp context)
- Add debug logging to silent exception handlers in proxy

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

proxy.py

@@ -599,8 +599,8 @@ async def lm_remote_proxy_middleware(request: web.Request, handler):
                 body = json.loads(response.body)
                 enriched = _enrich_list_response(body)
                 return web.json_response(enriched)
-            except Exception:
-                pass  # Return original response on any error
+            except Exception as exc:
+                logger.debug("[LM-Remote] Stats enrichment failed: %s", exc)
         # Inject stats UI script into HTML pages
         if (path in _PROXY_PAGE_ROUTES or path.rstrip("/") in _PROXY_PAGE_ROUTES):
             if response.status == 200:
@@ -616,8 +616,8 @@ async def lm_remote_proxy_middleware(request: web.Request, handler):
                             charset="utf-8",
                             status=response.status,
                         )
-                    except Exception:
-                        pass
+                    except Exception as exc:
+                        logger.debug("[LM-Remote] HTML injection failed: %s", exc)
         return response
 
     # Not a LoRA Manager route — fall through

stats_db.py

@@ -32,7 +32,7 @@ class StatsDB:
 
     def init(self) -> None:
         """Create the database and table if they don't exist."""
-        self._conn = sqlite3.connect(str(self._db_path))
+        self._conn = sqlite3.connect(str(self._db_path), check_same_thread=False)
         self._conn.row_factory = sqlite3.Row
         self._conn.execute("PRAGMA journal_mode=WAL")
         self._conn.executescript(_SCHEMA)

stats_service.py

@@ -72,25 +72,28 @@ class StatsFetchService:
             )
         return self._session
 
-    async def _fetch_model(self, model_id: int) -> dict | None:
+    async def _fetch_model(self, model_id: int, _retries: int = 2) -> dict | None:
         """Fetch a single model's data from CivitAI API."""
         url = f"{_CIVITAI_API}/models/{model_id}"
         session = await self._get_session()
-        try:
-            async with session.get(url) as resp:
-                if resp.status == 429:
-                    logger.warning("[LM-Remote] CivitAI rate limited, backing off")
-                    await asyncio.sleep(5)
-                    return None
-                if resp.status != 200:
-                    logger.debug("[LM-Remote] CivitAI returned %d for model %d",
-                                 resp.status, model_id)
-                    return None
-                return await resp.json()
-        except Exception as exc:
-            logger.warning("[LM-Remote] CivitAI fetch failed for model %d: %s",
-                           model_id, exc)
-            return None
+        for attempt in range(_retries + 1):
+            try:
+                async with session.get(url) as resp:
+                    if resp.status == 429:
+                        logger.warning("[LM-Remote] CivitAI rate limited, backing off (attempt %d)",
+                                       attempt + 1)
+                        await asyncio.sleep(5 * (attempt + 1))
+                        continue
+                    if resp.status != 200:
+                        logger.debug("[LM-Remote] CivitAI returned %d for model %d",
+                                     resp.status, model_id)
+                        return None
+                    return await resp.json()
+            except Exception as exc:
+                logger.warning("[LM-Remote] CivitAI fetch failed for model %d: %s",
+                               model_id, exc)
+                return None
+        return None
 
     async def fetch_stats_for_models(
         self,