Database/AI/ML frameworks & serving
JupyterHub: unauthenticated logins write unbounded usernames to the log, exhausting storage
Impact
A failed form login writes the attacker-supplied username to the hub log with no length bound, so anyone who can reach the login page can push arbitrary volume into the log stream at will. Where hub logs land on the node's root filesystem or are shipped to a metered logging backend, this is a cheap way to fill a disk or burn a quota. On a GPU cluster the hub is the front door for every user's notebook: if it wedges because the filesystem is full, no one's server spawns, and the host needs manual cleanup before the fleet is usable again. No confidentiality or integrity impact.
Who can reach it
Anyone who can reach the hub's login endpoint. No authentication and no valid account needed. The reachable population is whoever gets past whatever fronts the hub - internal-only exposure or an SSO proxy narrows it considerably.
What to do
Upgrade JupyterHub to 5.5.0 and restart the hub process. Until then, rate-limit the login endpoint at the proxy in front of the hub and confirm hub logs rotate with a hard size cap so a flood cannot fill the volume. The advisory names no other mitigation.
References
Related entries
- vLLM: malformed JSON to the OpenAI-compatible endpoints returns server paths and versionsCVE-2026-73555 · vLLM OpenAI-compatible API server (validation_exception_handler, sanitize_message)Medium
- vLLM: attacker-supplied structured-output regex pins a CPU core and stalls the engine pathCVE-2026-73556 · vLLM structured outputs, lm-format-enforcer backend (structured_outputs.regex)Medium
- vLLM: integer overflow in the activation CUDA kernel leaks another batched request's outputCVE-2026-73558 · vLLM CUDA activation kernels (act_and_mul_kernel, activation_kernels.cu)Medium
- vLLM (DeepStream video backend, VideoMediaIO backend selection): A performance feature merged past two existingNCVD-2026-044-vllm-deepstream-video-backend-vi · vLLM (DeepStream video backend, VideoMediaIO backend selection)Medium
- NVIDIA OpenShell for Linux: improper output encoding in the inference proxy leaks data and allows tamperingCVE-2026-65085 · NVIDIA OpenShell for Linux (inference proxy)Medium
- Pure Storage FlashArray key rotation logging (Rapid Data Locking): The Key Encryption Key is written to logs duringCVE-2025-2327 · Pure Storage FlashArray key rotation logging (Rapid Data Locking)Medium
This entry is curated: imported from vendor advisories with machine assistance, not yet individually verified. Confirm against your vendor's advisory before acting, and report anything wrong.