The aiohttp server used in Scrapy versions from 2.19.0 up to but not including 2.19.0_2 lacks an initial request/header timeout, allowing an attacker to keep an idle connection open indefinitely before sending any data. This can lead to resource exhaustion. A fix is available in version 2.19.0_2. The recommended mitigation is to use a reverse proxy with sensible keepalive limits, which is advised for all production deployments.