SSRF via X-Request-Id header drives P2pNcclConnector outbound ZMQ+NCCL pair, leaking KV-cache tensors
GHSA-vfp2-c8pq-v6h6 · published 2026-10-09
- Severity
- High
- CVSS score
- Not scored
- Chance of exploitation (EPSS)
- No CVE
- Fixed in
- 0.24.0
Any API client that can reach a prefill worker running P2pNcclConnector can exfiltrate the full KV-cache (prompt text and prefix-cached hidden states) of any request that worker processes, using only an HTTP header. The leak also includes the NCCL communicator unique_id, and the blocking ncclCommInitRank call against an unresponsive attacker host constitutes a denial-of-service vector against the prefill worker. CWE-918 (Server-Side Request Forgery), severity high.
Upgrade vllm to 0.24.0 or later. If you start it from an MCP config with npx or uvx, pin that version so an old cached copy is not used.
Package: vllm (PyPI). Affected versions: >= 0.9.2, < 0.24.0, when using P2pNcclConnector. Patched version: 0.24.0. PR #44854 removed the affected connector and its registration. OpenAIServing._base_request_id in must strip or replace X-Request-Id with a random UUID when P2pNcclConnector is active, preventing any user-supplied value from reaching parse_request_id. Additionally, create_connect in must validate remote_address against an operator-configured CIDR/hostname allow-list (supplied via kv_connector_extra_config) before opening the ZMQ socket and initiating NCCL bootstrap.
| Package | Registry | Affected | Fixed in |
|---|---|---|---|
| vllm | PyPI | >= 0.9.2, < 0.24.0 | 0.24.0 |
- CriticalMLflow: Unauthenticated full-read SSRF in webhook delivery: _validate_webhook_url bypassed via unvalidated HTTP redirects (and DNS rebindingmlflowExploited
- Critical@zereight/mcp-gitlab Vulnerable to Server-Side Request Forgery@zereight/mcp-gitlab
- CriticalSSRF + LLM API key exfiltration via user-controlled basepath / baseUrl in 8 chat-model nodesflowise, flowise-components
- CriticalLocalAI POST /models/apply permits unauthenticated server-side request forgery through gallery URLsgithub.com/mudler/LocalAI
- CriticalFastMCP OpenAPI Provider has an SSRF & Path Traversal Vulnerabilityfastmcp
- CriticalBentoML SSRF Vulnerability in File Upload Processingbentoml
- CriticalvLLM: OpenAI auth bypassvllm
- CriticalvLLM has RCE In Video Processingvllm
- CriticalvLLM Allows Remote Code Execution via PyNcclPipe Communication Servicevllm
- CriticalvLLM Vulnerable to Remote Code Execution via Mooncake Integrationvllm
- CriticalCVE-2025-24357 Malicious model remote code execution fix bypass with PyTorch < 2.6.0vllm
- CriticalvLLM Deserialization of Untrusted Data vulnerabilityvllm
The impact, patch and workaround text comes from the advisory, shortened (published by vLLM). Proof-of-concept and reproduction sections and code are left out; the original advisory has them.