feat: graceful shutdown and WAL cleanup

- Add stop_grace_period to backend container for graceful shutdown
- Document WAL mode rationale and orphaned file cleanup in db.py
- Handle database close errors gracefully in lifespan
- Clean up orphaned WAL files during startup before opening DB
- Reorder imports and fix formatting in startup.py

Docker/compose.prod.yml

@@ -48,6 +48,7 @@ services:
       target: runtime
     container_name: bangui-backend
     restart: unless-stopped
+    stop_grace_period: 30s  # Give lifespan 30s to complete before SIGKILL
     depends_on:
       fail2ban:
         condition: service_healthy

backend/app/db.py

@@ -274,7 +274,18 @@ CREATE INDEX IF NOT EXISTS idx_import_log_source_id_desc
 
 
 async def _configure_connection(db: aiosqlite.Connection) -> None:
-    """Apply hardening pragmas to a newly-opened SQLite connection."""
+    """Apply hardening pragmas to a newly-opened SQLite connection.
+
+    WAL mode is intentionally kept despite the risk of orphaned ``.wal``/``.shm``
+    files after unclean shutdowns.  The benefits for concurrent readers
+    (readers do not block writers) outweigh the cleanup overhead, especially
+    under load.  BanGUI runs as a single worker, but multiple concurrent HTTP
+    requests can still issue overlapping reads; DELETE mode would serialize
+    those reads behind any write, degrading API performance.
+
+    Orphaned files are handled by :func:`_cleanup_wal_files`, which is called
+    during startup before the database is opened.
+    """
     await db.execute("PRAGMA journal_mode=WAL;")
     await db.execute("PRAGMA foreign_keys=ON;")
     await db.execute("PRAGMA busy_timeout=5000;")

backend/app/main.py

@@ -318,7 +318,12 @@ async def _lifespan(app: FastAPI) -> AsyncGenerator[None, None]:
             log.error("scheduler_lock_release_failed", error=str(e))
 
         # 6. Close the database connection.
-        await startup_db.close()
+        try:
+            await startup_db.close()
+            log.debug("database_connection_closed")
+        except Exception as exc:
+            log.error("database_connection_close_failed", error=str(exc))
+
         log.info("bangui_shut_down")
 
 

backend/app/startup.py

@@ -26,10 +26,9 @@ from pathlib import Path
 from typing import TYPE_CHECKING, Any
 
 import aiohttp
-from app.utils.logging_compat import get_logger
 from apscheduler.schedulers.asyncio import AsyncIOScheduler  # type: ignore[import-untyped]
 
-from app.db import init_db, open_db
+from app.db import _cleanup_wal_files, init_db, open_db
 from app.services import setup_service
 from app.services.dns_validated_connector import create_dns_validated_socket_factory
 from app.services.geo_cache import GeoCache
@@ -48,6 +47,7 @@ from app.tasks import (
 from app.utils.async_utils import run_blocking
 from app.utils.fail2ban_db_utils import ensure_fail2ban_indexes
 from app.utils.jail_config import ensure_jail_configs
+from app.utils.logging_compat import get_logger
 from app.utils.runtime_state import set_runtime_settings
 from app.utils.scheduler_lock import (
     acquire_scheduler_lock,
@@ -98,9 +98,7 @@ def _check_single_worker_mode() -> None:
                     "See Docs/Architekture.md § Deployment Constraints for details."
                 )
         except ValueError as e:
-            raise RuntimeError(
+            raise RuntimeError(f"BANGUI_WORKERS environment variable must be an integer, got: {workers_env}") from e
-                f"BANGUI_WORKERS environment variable must be an integer, got: {workers_env}"
-            ) from e
 
 
 async def _ensure_database_schema(database_path: str) -> None:
@@ -333,6 +331,11 @@ async def _stage_init_database(app: FastAPI, settings: Settings) -> Any:
 
     log.debug("database_directory_ensured", directory=str(db_path.parent))
 
+    # Clean up orphaned WAL files from previous unclean shutdowns before
+    # opening the database.  This prevents stale .wal/.shm files from
+    # interfering with startup or triggering misleading warnings.
+    await _cleanup_wal_files(settings.database_path)
+
     original_db_path = db_path.resolve()
     startup_db = await open_db(settings.database_path)
 
@@ -357,9 +360,7 @@ async def _stage_init_database(app: FastAPI, settings: Settings) -> Any:
                     if f2b_db_path:
                         await run_blocking(ensure_fail2ban_indexes, f2b_db_path)
 
-                    persisted_runtime_settings = (
+                    persisted_runtime_settings = await setup_service.get_persisted_runtime_settings(runtime_db)
-                        await setup_service.get_persisted_runtime_settings(runtime_db)
-                    )
                 finally:
                     await runtime_db.close()