Your esalao-queue service is crashing with repeated connection timeouts to redis.railway.internal:6379 via Predis, while your Redis service in the same project is running normally. If your environment was created before October 16, 2025, private networking is IPv6-only, and Predis using tcp:// may default to IPv4 resolution, which would fail. You can verify this by temporarily switching to the public Redis URL (REDIS_PUBLIC_URL), and if that works, you'll need to configure Predis to connect over IPv6 or use the public URL. See our private networking docs for details on legacy environments.