fix: add SQLite connection parameters to ND_DBPATH and wait for DB file

entrypoint.sh

@@ -70,9 +70,25 @@ start_app() {
   # Ensure shared directories exist on persistent host volume
   mkdir -p /data/plugins /data/cache /data/backup
 
+  # Explicitly wait for the DB file to appear in the LiteFS mount
+  local db_file="/litefs/navidrome.db"
+  local timeout=30
+  local count=0
+  
+  echo "Waiting for LiteFS database at ${db_file}..."
+  while [ ! -f "$db_file" ] && [ $count -lt $timeout ]; do
+    sleep 1
+    count=$((count + 1))
+  done
+
+  if [ ! -f "$db_file" ]; then
+     echo "WARNING: Database file ${db_file} not found after ${timeout}s. LiteFS may still be initializing."
+     # We continue anyway as Navidrome might create it, but logging this is vital for debugging.
+  fi
+
   # Tell Navidrome to use the database directly from the LiteFS mount.
-  # This avoids symlink issues on filesystems that don't support them.
-  export ND_DBPATH="/litefs/navidrome.db"
+  # We include standard Navidrome SQLite params for stability.
+  export ND_DBPATH="${db_file}?cache=shared&_busy_timeout=15000&_journal_mode=WAL&_foreign_keys=on"
 
   /app/navidrome &
   NAVIDROME_PID=$!