vllm p2p never connected zmq_exhaustion

[ Description ]Official description

vLLM through 0.29.0 contains a denial of service vulnerability in P2P KV offloading when OffloadingConnector is configured with TieringOffloadingSpec and a peer-to-peer secondary tier. Attackers can supply arbitrary remote host and port values in kv_transfer_params to create unreachable peer sessions that retain ZeroMQ sockets until the context quota is exhausted, causing an uncaught ZMQError that crashes EngineCore and stops all inference.

[ Fixes ]Fixes and patches

• vllm 0.32.0Fixed version · GitHub Advisories · pip package vllm: upgrade to 0.32.0 or later (affected < 0.32.0)

[ Packages ]Affected open-source packages

• vllm pipAffected < 0.32.0 · fixed in 0.32.0 · GHSA-frvg-qw88-hww9

[ References ]References

• github.com/vllm-project/vllm/pull/51504issue-trackingpatch
• github.com/vllm-project/vllm/blob/v0.29.0/vllm/v1/kv_offload/tiering/p2p/control/zmq.pytechnical-description
• github.com/vllm-project/vllm/blob/v0.29.0/vllm/v1/kv_offload/tiering/p2p/manager.pytechnical-description
• github.com/vllm-project/vllmproduct
• www.vulncheck.com/advisories/vllm-through-0.29.0-denial-of-service-via-unbounded-p2p-kv…third-party-advisory