← Voltar ao feed
high

Critical Rancher Fleet Flaws Allow Privilege Escalation to Cluster-Admin (CVE-2026-94290, CVE-2026-104615)

1 fonte(s) · 1 atualização(ões) · publicado 03/11/2026 21:00 · coletado há 1 hora

SUSE released security advisories on October 29, 2026 addressing two serious vulnerabilities in Rancher Fleet, the GitOps engine used for multi-cluster Kubernetes management. The most critical, CVE-2026-94290 (CVSSv3 9.1), allows an authenticated Rancher user with only default privileges to escalate to full cluster-admin on the management cluster by referencing administrative kubeconfig secrets in a custom Fleet Cluster object. A second flaw, CVE-2026-104615, enables cluster identity takeover during registration, letting an attacker holding a valid workspace token claim another cluster's identity and access its secrets. No active exploitation or public PoC has been reported, but patches are available and should be applied promptly given the severity of the privilege escalation risk.

Entenda a falha

A vulnerabilidade

CVE-2026-94290 (CVSSv3 9.1) is a privilege escalation in the agent management controller of Rancher Fleet, the GitOps engine for multi-cluster Kubernetes. An authenticated Rancher user holding only the default user global role, with no role bindings on the local cluster, can become cluster-admin on the Rancher management cluster. CVE-2026-104615 is an improper authorization flaw in cluster registration: an attacker with a valid workspace token can claim another cluster's identity, reach its secrets and disrupt management services. Affected: standalone Fleet before 0.16.3, 0.15.8, 0.14.12 and 0.13.17, and Fleet bundled in Rancher before 2.15.3, 2.14.7, 2.13.11 and 2.12.15. Fixed in those versions or later.

Como funciona

For CVE-2026-94290, a user creates a custom Fleet Cluster object that points to a kubeconfig secret. Fleet's agent management controller reads that object and uses the referenced kubeconfig to act on the cluster. The controller runs with cluster-admin rights on the management cluster, so if the object references an administrative kubeconfig secret, the controller's actions run with that authority. For CVE-2026-104615, a cluster sends a registration request that states which cluster identity it claims and presents a registration token. The server accepted the claimed identity without checking that it matched the token used.

Por que é vulnerável

Both are missing authorization checks. In CVE-2026-94290 the controller did not sufficiently restrict which kubeconfig secrets a Fleet Cluster object could reference, while running with cluster-admin privileges, which is a confused-deputy situation. In CVE-2026-104615 the cluster identity claimed in a registration request was not bound to the registration token used for the request. The articles do not say why they went unnoticed. The fixes in the patched releases address these missing restrictions and binding.

Onde está

The articles name the Fleet agent management controller for CVE-2026-94290 and the cluster registration flow for CVE-2026-104615, with advisories GHSA-qfmw-gw96-ffcr and GHSA-gx8v-2929-9q95; no files or functions are named.

Como foi descoberta

The articles do not say who found the flaws or how. SUSE published security advisories on October 29, 2026, with GitHub advisories GHSA-qfmw-gw96-ffcr and GHSA-gx8v-2929-9q95 describing them. SUSE reports no exploitation in the wild so far.

Explicação gerada por IA a partir da matéria e dos artigos, em 04/11/2026 09:10.

Detalhes técnicos

CVE-2026-94290 stems from Fleet's agent management controller failing to restrict which kubeconfig secrets a Fleet Cluster object could reference; since the controller runs with cluster-admin privileges, an attacker-controlled Cluster object executes with root-level authority. CVE-2026-104615 arises because a cluster identity claimed during a registration request was not properly bound to the registration token used, allowing token holders to impersonate other clusters' identities and disrupt management services.

Remediação

Update standalone Fleet to versions 0.16.3, 0.15.8, 0.14.12, or 0.13.17 (or later), or update bundled Rancher to 2.15.3, 2.14.7, 2.13.11, or 2.12.15 (or later). Administrators should also remove workspace self-service rights from untrusted accounts and audit clusters for unexpected role bindings.

Linha do tempo

  1. securityonline.info ↗ · pub. 03/11/2026 21:00 · coletado 04/11 08:52 primeiro registro

    Critical Rancher Fleet Flaws Allow Privilege Escalation to Cluster-Admin (CVE-2026-94290, CVE-2026-104615)

    ler a análise desta fonte

    SUSE released security advisories on October 29, 2026 addressing two serious vulnerabilities in Rancher Fleet, the GitOps engine used for multi-cluster Kubernetes management. The most critical, CVE-2026-94290 (CVSSv3 9.1), allows an authenticated Rancher user with only default privileges to escalate to full cluster-admin on the management cluster by referencing administrative kubeconfig secrets in a custom Fleet Cluster object. A second flaw, CVE-2026-104615, enables cluster identity takeover during registration, letting an attacker holding a valid workspace token claim another cluster's identity and access its secrets. No active exploitation or public PoC has been reported, but patches are available and should be applied promptly given the severity of the privilege escalation risk.

    Technical Details

    CVE-2026-94290 stems from Fleet's agent management controller failing to restrict which kubeconfig secrets a Fleet Cluster object could reference; since the controller runs with cluster-admin privileges, an attacker-controlled Cluster object executes with root-level authority. CVE-2026-104615 arises because a cluster identity claimed during a registration request was not properly bound to the registration token used, allowing token holders to impersonate other clusters' identities and disrupt management services.

    Remediation

    Update standalone Fleet to versions 0.16.3, 0.15.8, 0.14.12, or 0.13.17 (or later), or update bundled Rancher to 2.15.3, 2.14.7, 2.13.11, or 2.12.15 (or later). Administrators should also remove workspace self-service rights from untrusted accounts and audit clusters for unexpected role bindings.