CVE-2026-44939: Rancher Manager RCE Vulnerability

CVE-2026-44939 Overview

CVE-2026-44939 is a command injection vulnerability in Rancher Manager versions before 2.14.2. The flaw resides in the cluster import endpoint /v3/import/{token}_{clusterId}.yaml, which fails to sanitize YAML parameters before processing. Remote attackers can leverage this weakness to break out of a container image and execute malicious containers within the managed cluster. The vulnerability is categorized under [CWE-95] Improper Neutralization of Directives in Dynamically Evaluated Code. It carries a CVSS 4.0 score of 9.4 and impacts both confidentiality and integrity of the entire Kubernetes management plane.

Critical Impact
Unauthenticated remote attackers can inject commands through unsanitized YAML parameters and execute arbitrary containers within Rancher-managed Kubernetes clusters.

Affected Products

Rancher Manager versions prior to 2.14.2
Rancher Manager cluster import endpoint /v3/import/{token}_{clusterId}.yaml
Kubernetes clusters managed by vulnerable Rancher Manager instances

Discovery Timeline

2026-06-19 - CVE-2026-44939 published to NVD
2026-06-24 - Last updated in NVD database

Technical Details for CVE-2026-44939

Vulnerability Analysis

The vulnerability exists in the Rancher Manager cluster import workflow. The endpoint /v3/import/{token}_{clusterId}.yaml accepts YAML content used to bootstrap new clusters into Rancher management. The handler processes YAML parameters without applying sufficient input validation or sanitization before evaluation. Attackers craft YAML payloads that inject directives interpreted as executable instructions during the import process. Successful exploitation allows the attacker to break out of the expected image context and launch arbitrary containers on the target cluster. The scope change reflected in the CVSS 4.0 vector indicates downstream impact on managed cluster resources beyond the Rancher Manager itself.

Root Cause

The root cause is improper neutralization of directives in dynamically evaluated code [CWE-95]. The import endpoint trusts user-supplied YAML parameters and uses them in a context where they are interpreted as commands rather than data. The absence of strict schema validation, allowlisting, and sandboxing on the YAML processor enables the injection.

Attack Vector

The attack vector is network-based and requires user interaction to import a malicious cluster manifest. An attacker supplies a crafted YAML document to the /v3/import/{token}_{clusterId}.yaml endpoint. When processed by Rancher Manager, the injected directives execute, enabling the attacker to deploy malicious containers in the managed cluster. Refer to the Rancher GitHub Security Advisory GHSA-mhc6-2gfq-xx62 for full technical context.

Detection Methods for CVE-2026-44939

Indicators of Compromise

Unexpected POST or PUT requests against /v3/import/{token}_{clusterId}.yaml originating from untrusted networks.
Unscheduled container or pod deployments appearing in managed clusters shortly after import API activity.
YAML payloads to the import endpoint containing shell metacharacters, embedded scripts, or unusual command/args fields.
Outbound network connections from newly created containers to unknown external endpoints.

Detection Strategies

Inspect Rancher Manager audit logs for requests to the cluster import endpoint and correlate with subsequent cluster state changes.
Apply web application firewall rules that flag YAML parameters containing executable directives or shell metacharacters at the import endpoint.
Monitor Kubernetes audit logs for pod creation events lacking a corresponding administrative ticket or change request.

Monitoring Recommendations

Forward Rancher API access logs and Kubernetes audit logs to a centralized SIEM for correlation.
Baseline normal cluster import frequency and alert on volume anomalies or off-hours activity.
Track container image provenance and alert on images pulled from unapproved registries.

How to Mitigate CVE-2026-44939

Immediate Actions Required

Upgrade Rancher Manager to version 2.14.2 or later without delay.
Restrict network access to the Rancher Manager API to trusted administrative networks only.
Audit recent cluster import activity and inspect managed clusters for unauthorized workloads.
Rotate cluster registration tokens after patching to invalidate any leaked credentials.

Patch Information

Rancher addressed the issue in Rancher Manager 2.14.2. Administrators should apply the upgrade following standard Rancher procedures. Full remediation details are available in the Rancher GitHub Security Advisory GHSA-mhc6-2gfq-xx62.

Workarounds

Block external access to the /v3/import/ endpoint at the reverse proxy or ingress layer until patching is complete.
Enforce strict role-based access control on cluster import operations and limit which accounts can submit YAML manifests.
Deploy admission controllers in managed clusters to reject pods that lack approved image signatures or originate from unapproved registries.

bash

# Example ingress restriction limiting access to the Rancher import endpoint
# Place behind an authenticated reverse proxy and allowlist admin networks
location ~ ^/v3/import/ {
    allow 10.0.0.0/24;   # admin subnet
    deny all;
    proxy_pass https://rancher-backend;
}