Skip to main content
CVE Vulnerability Database
Vulnerability Database/CVE-2024-10110

CVE-2024-10110: Aimstack Aim DOS Vulnerability

CVE-2024-10110 is a denial of service vulnerability in Aimstack Aim that blocks the main thread indefinitely, preventing the tracking server from responding. This article covers technical details, affected versions, and mitigation.

Published:

CVE-2024-10110 Overview

CVE-2024-10110 is a denial of service vulnerability affecting aimhubio/aim version 3.23.0. The flaw resides in the ScheduledStatusReporter object, which can be instantiated to run on the main thread of the tracking server. When this happens, the main thread becomes blocked indefinitely, preventing the tracking server from responding to any subsequent requests. The vulnerability is categorized under CWE-400 (Uncontrolled Resource Consumption) and is exploitable over the network without authentication or user interaction.

Critical Impact

Remote attackers can render the Aim tracking server unresponsive by triggering ScheduledStatusReporter instantiation on the main thread, halting all experiment tracking operations.

Affected Products

  • aimstack aim 3.23.0 (Python package)
  • Deployments using the Aim tracking server for ML experiment tracking
  • Environments referencing cpe:2.3:a:aimstack:aim:3.23.0:*:*:*:*:python:*:*

Discovery Timeline

  • 2025-03-20 - CVE-2024-10110 published to NVD
  • 2026-06-17 - Last updated in NVD database

Technical Details for CVE-2024-10110

Vulnerability Analysis

Aim is an open-source experiment tracking tool used by machine learning teams to log and compare training runs. The tracking server exposes a network-accessible API for ingesting metrics and metadata from remote clients. In version 3.23.0, the ScheduledStatusReporter component performs periodic status reporting through a blocking loop. When this reporter is instantiated on the server's main thread rather than a dedicated worker thread, the main event loop is prevented from processing further inbound requests.

The result is a complete denial of service: connected clients receive no responses, new connections cannot be established, and pending experiment tracking data cannot be committed. Because the attacker requires no privileges and no user interaction to trigger the condition, exposure of the tracking server to untrusted networks introduces meaningful availability risk.

Root Cause

The root cause is improper thread management in the ScheduledStatusReporter lifecycle. The component runs its scheduled work synchronously on whichever thread instantiates it. When that thread is the main request-handling thread, the scheduler monopolizes execution and starves other server operations. This maps directly to CWE-400, where an actor consumes resources in a manner that the server cannot bound or reclaim.

Attack Vector

An unauthenticated remote attacker with network reachability to the Aim tracking server can trigger the condition through interactions that cause ScheduledStatusReporter to be instantiated on the main thread. Once triggered, the tracking server becomes unresponsive until it is restarted. Refer to the Huntr Bounty Listing for the technical write-up documenting the exploit pathway.

No verified public proof-of-concept code has been published in the enriched dataset, so the exploitation mechanism is described here in prose only.

Detection Methods for CVE-2024-10110

Indicators of Compromise

  • Aim tracking server processes that stop responding to HTTP requests while remaining alive in the process table.
  • Client-side timeouts and hung connections from experiment tracking SDKs targeting the Aim server endpoint.
  • Sudden cessation of new experiment records being persisted to the Aim backend storage.

Detection Strategies

  • Inventory Python environments and container images for aim==3.23.0 using package manifests, SBOMs, or pip freeze output.
  • Monitor the tracking server's main thread with health-check endpoints that return within a defined latency budget.
  • Correlate network telemetry for repeated failed connections to the Aim server port against process liveness data.

Monitoring Recommendations

  • Configure external synthetic probes against the Aim tracking API and alert when response latency exceeds baseline thresholds.
  • Capture process-level thread state metrics to identify main-thread blocking patterns.
  • Log and review all inbound requests that immediately precede a server hang for forensic reconstruction.

How to Mitigate CVE-2024-10110

Immediate Actions Required

  • Restrict network exposure of the Aim tracking server to trusted management networks or place it behind an authenticated reverse proxy.
  • Upgrade aimhubio/aim to a release later than 3.23.0 once a fixed version is confirmed by the maintainers.
  • Implement automated restart supervision so a hung tracking server is recycled quickly to restore availability.

Patch Information

At the time of publication, the enriched CVE record does not list a vendor advisory URL or a specific fixed version. Consult the Huntr Bounty Listing and the upstream aimhubio/aim repository for the latest remediation status before deploying a fix.

Workarounds

  • Disable or avoid instantiating ScheduledStatusReporter on the main thread; run scheduled reporting in a dedicated worker thread or process.
  • Place the Aim tracking server behind a network policy that limits access to authenticated ML platform users only.
  • Apply request rate limiting and connection timeouts at a reverse proxy to reduce the window for triggering the blocking condition.
bash
# Configuration example: restrict Aim tracking server exposure with a firewall rule
# Allow only the ML platform subnet to reach the Aim server port
iptables -A INPUT -p tcp --dport 43800 -s 10.10.0.0/24 -j ACCEPT
iptables -A INPUT -p tcp --dport 43800 -j DROP

Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

Default Legacy - Prefooter | Experience the World’s Most Advanced Cybersecurity Platform

Experience the Most Advanced Cybersecurity Platform

See how the world’s most intelligent, autonomous cybersecurity platform can protect your organization today and into the future.