CVE-2025-30202: vLLM: ZeroMQ socket exposure enables DoS in multi-node

vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Versions starting from 0.5.2 and prior to 0.8.5 are vulnerable to denial of service and data exposure via ZeroMQ on multi-node vLLM deployment. In a multi-node vLLM deployment, vLLM uses ZeroMQ for some multi-node communication purposes. The primary vLLM host opens an XPUB ZeroMQ socket and binds it to ALL interfaces. While the socket is always opened for a multi-node deployment, it is only used when doing tensor parallelism across multiple hosts. Any client with network access to this host can connect to this XPUB socket unless its port is blocked by a firewall. Once connected, these arbitrary clients will receive all of the same data broadcasted to all of the secondary vLLM hosts. This data is internal vLLM state information that is not useful to an attacker. By potentially connecting to this socket many times and not reading data published to them, an attacker can also cause a denial of service by slowing down or potentially blocking the publisher. This issue has been patched in version 0.8.5.

An attacker with network access to a corporate AI inference cluster—via a compromised internal host or misconfigured network segment—port-scans and discovers the ZeroMQ XPUB socket open on port 5557 of the primary vLLM node. Using a trivial ZeroMQ subscriber script (under 10 lines of Python), the attacker spawns hundreds of simultaneous connections that subscribe but never consume data. The publisher's send buffer fills; internal state broadcasts to legitimate secondary nodes slow and eventually block entirely. The multi-node LLM inference cluster degrades then stalls, taking down all AI-powered applications relying on that serving layer. No credentials, no malware, no AI knowledge required—just a ZeroMQ client library.