CVE-2026-118761

Kubeflow Pipelines: Pipeline archive upload allows decompression bomb to exhaust server memory

Published Oct 29, 2026 CVSS 3.1 · 7.5 High EPSS < 1% KEV · Not listed
Description
Kubeflow Pipelines enables users to build and deploy portable, scalable machine learning workflows. Prior to 2.18.0, the API server decompressed uploaded pipeline archives without adequately bounding expanded data while locating or reading the pipeline file. A small ZIP or gzip-compressed tar archive can consume disproportionately large memory and exhaust API-server resources. Multipart pipeline and pipeline-version upload processing performs decompression before namespace authorization, so a reachable caller does not need a successful pipeline-create authorization decision. URL-import handlers authorize before fetching and decompressing a remote archive. The amount of expansion required to interrupt service depends on deployment resource limits and concurrent activity, and no specific archive is guaranteed to crash every deployment. This issue is fixed in version 2.18.0.

Analysis

AI-generated

Technical impact

The vulnerability arises from the Kubeflow Pipelines API server decompressing pipeline archive uploads without size or expansion limits. When a maliciously crafted ZIP or gzip archive is uploaded, the server attempts to decompress the entire archive before verifying namespace authorization or pipeline existence. This can lead to a very large memory footprint, potentially exhausting API‑server resources and causing it to become unresponsive. The flaw is a classic resource‑exhaustion flaw and is identified as CWE‑409.

Affected Systems

The flaw affects Kubernetes deployments running Kubeflow Pipelines prior to version 2.18.0. The impacted component is the API server responsible for pipeline and pipeline‑version uploads, including multipart and URL‑import handlers. Any cluster using an older Pipelines release is vulnerable until the component is updated.

Risk and Exploitability

The CVSS score of 7.5 reflects a high risk, combined with the lack of required authentication for upload before decompression. An attacker who can reach the API endpoint, whether they have valid credentials or not, can trigger the archive decompression. The absence of EPSS data does not remove the concern; a decompression bomb is a well‑known attack technique that can lead to denial of service if the application fails to enforce size limits. Since the vulnerability is in the core upload logic and does not depend on additional configuration, it is likely exploitable in many deployments.

Remediation

No solution or workaround provided in the CVE record.

OpenCVE Recommended Actions

  • Upgrade Kubeflow Pipelines to version 2.18.0 or later, where the fixed logic limits decompression expansion.
  • If an immediate upgrade is not possible, enforce strict size limits on uploads at the API gateway or reverse proxy and apply tight Kubernetes resource limits to the API‑server pod.
  • Restrict the API server to run under a dedicated service account with the minimal necessary permissions and consider isolating it from untrusted traffic via network segmentation.
  • Monitor API‑server memory usage and set alerts for sudden spikes to detect potential exploitation.
  • Validate or scan uploaded archives for acceptable size before initiating decompression as an additional defensive measure.

Generated by OpenCVE AI on October 29, 2026 at 17:27 UTC.

Tracking

Sign in to view the affected projects.

Advisories

Source ID Title
Github GHSA Github GHSA GHSA-8fmv-6rg5-5924 Kubeflow Pipelines: Pipeline archive upload allows decompression bomb to exhaust server memory
History
Type Values Removed Values Added
Metrics ssvc

{'options': {'Automatable': 'yes', 'Exploitation': 'none', 'Technical Impact': 'partial'}, 'version': '2.0.3'}

Type Values Removed Values Added
First Time appeared Kubeflow
Kubeflow pipelines
Vendors & Products Kubeflow
Kubeflow pipelines
Type Values Removed Values Added
Description Kubeflow Pipelines enables users to build and deploy portable, scalable machine learning workflows. Prior to 2.18.0, the API server decompressed uploaded pipeline archives without adequately bounding expanded data while locating or reading the pipeline file. A small ZIP or gzip-compressed tar archive can consume disproportionately large memory and exhaust API-server resources. Multipart pipeline and pipeline-version upload processing performs decompression before namespace authorization, so a reachable caller does not need a successful pipeline-create authorization decision. URL-import handlers authorize before fetching and decompressing a remote archive. The amount of expansion required to interrupt service depends on deployment resource limits and concurrent activity, and no specific archive is guaranteed to crash every deployment. This issue is fixed in version 2.18.0.
Title Kubeflow Pipelines: Pipeline archive upload allows decompression bomb to exhaust server memory
Weaknesses CWE-409
References
Metrics cvssV3_1

{'score': 7.5, 'vector': 'CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H'}

Weaknesses
  • CWE-409

    Improper Handling of Highly Compressed Data (Data Amplification)

Sources

Status
PUBLISHED
Assigner
GitHub_M
Published
Updated
Oct 29, 2026, 8:18 PM
Reserved
Oct 28, 2026, 7:20 PM
Status
Awaiting Analysis
Published
Oct 29, 2026, 5:18 PM
Modified
Oct 30, 2026, 5:18 PM
Severity
Important
Public date
Oct 29, 2026, 4:37 PM