/

CVE-2026-44543

Severity:

HIGH RISK

(8.7)

Privilege Management / Escalation in Rancher Local Path

CVE-2026-44543

Analyzed

Vulnerability type

Category:

Privilege Management / Escalation

CWE:

CWE-269

Impact:

Privilege Escalation and Host Node Compromise

Severity:

HIGH RISK

(8.7)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

A privilege management flaw in Rancher Local Path Provisioner allows a user who can edit the local-path-config ConfigMap to inject security-sensitive fields into the helperPod.yaml template used for PVC operations.

When the provisioner creates HelperPods from this attacker-controlled template, it may start privileged pods with hostPath mounts, enabling broad access to the underlying Kubernetes node and co-located workloads.

Affected context

This vulnerability affects Kubernetes clusters that:

  • Deploy rancher/local-path-provisioner as a storage provisioner
  • Use the local-path-config ConfigMap in the local-path-storage namespace to define helperPod.yaml
  • Allow some users, controllers, or automation to edit this ConfigMap

The risk materializes when PersistentVolumeClaim (PVC) creation or cleanup triggers HelperPod creation based on the unvalidated template. Clusters with broad ConfigMap edit permissions or shared namespaces are particularly exposed.

Exploit preconditions

  • Attack surface: Kubernetes API (configuration layer)
  • Required access: Ability to edit or update the local-path-config ConfigMap in the local-path-storage namespace
  • Trigger point: Any PVC provisioning or cleanup that causes the Local Path Provisioner to instantiate a HelperPod from helperPod.yaml
  • Why exploitation is possible:
  • The provisioner trusts and directly applies the helperPod.yaml template without enforcing strict validation or policy checks on security-sensitive fields
  • This enables injection of securityContext.privileged, hostPath volumes, and Linux capabilities into pods created automatically by the provisioner

Why this happens

The core issue is improper privilege management and trust in user-editable configuration that controls pod specifications.

The Local Path Provisioner loads helperPod.yaml from a ConfigMap and uses it to create HelperPods without sanitizing or constraining fields that affect pod privilege, such as:

  • securityContext.privileged
  • hostPath volume definitions
  • Extended Linux capabilities

This aligns with CWE-269: Improper Privilege Management, where a component enables escalation of privileges by delegating security-critical configuration to less-trusted principals. The provisioner effectively acts as a privileged orchestrator that will instantiate high-privilege pods based on attacker-controlled templates.

Potential impact

  • Attacker-controlled creation of privileged pods on specific nodes
  • Mounting of the host root filesystem into a pod, enabling direct access to host files
  • Reading ServiceAccount tokens and secrets from other pods scheduled on the same node
  • Accessing other tenants' local-path volume data stored on the node
  • Modifying files on the host node, including binaries, configuration, or container runtime data
  • Establishing persistent backdoors at the node level or facilitating cluster-wide compromise through lateral movement

General mitigation guidance

Effective mitigation centers on reducing trust in user-controlled templates and tightening privilege boundaries around storage provisioners:

  • Constrain who can edit the local-path-config ConfigMap using fine-grained RBAC; treat it as a highly sensitive configuration object
  • Ensure that only tightly controlled automation or cluster administrators can alter pod templates used by storage provisioners
  • Apply admission controls or policies (e.g., PodSecurity standards, policy engines) that reject pods with disallowed fields such as:
  • securityContext.privileged: true
  • Arbitrary hostPath mounts, especially / or other sensitive paths
  • Elevated capabilities or host-level features
  • Prefer secure-by-default templates that omit privileged fields entirely and rely on minimal filesystem and capability requirements
  • Run the provisioner with least-privilege RBAC and avoid granting it unnecessary permissions beyond what is required for PVC provisioning

How teams detect issues like this

Organizations can detect and monitor for this pattern using a combination of configuration review, policy enforcement, and runtime monitoring:

  • Periodically audit the local-path-config ConfigMap for:
  • Presence of securityContext.privileged
  • hostPath volume usage, especially mounting / or other host system directories
  • Non-standard or broad Linux capabilities
  • Use admission controllers or policy engines to:
  • Enforce PodSecurity or custom policies on any pod created by storage provisioners
  • Block privileged pods or hostPath volumes except in tightly controlled namespaces
  • Implement configuration-as-code scanning in CI pipelines to flag dangerous changes to provisioner-related ConfigMaps and templates
  • Monitor cluster logs and events for:
  • New HelperPods created by the Local Path Provisioner with elevated privileges
  • Unusual hostPath mount patterns or privileged pod launches in the local-path-storage namespace

Sources

SourceLink
github.comgithub.com/rancher/local-path-provisioner/security/advisories/GHSA-7fxv-8wr2-mfc4

Frequently asked

questions

Not directly. Exploitation requires Kubernetes API access with permission to modify the local-path-config ConfigMap. However, any compromised in-cluster identity with those permissions can use this issue to escalate to node-level access.

The attacker needs sufficient RBAC privileges to edit or update the local-path-config ConfigMap in the local-path-storage namespace. Typical read-only or basic workload deployment permissions are not enough by themselves.

Treat pod templates used by system components as high-trust assets: restrict who can change them, and enforce policy that strips or blocks security-sensitive options (privileged mode, hostPath, capabilities) regardless of template contents. Combine this with least-privilege RBAC for the provisioner.

Avoid designing privileged controllers that blindly apply user-editable pod specs. Instead, separate user-configurable options from security-critical fields, validate configuration rigorously, and enforce cluster-wide policies that prevent privileged pods and unsafe hostPath mounts except where explicitly justified and tightly controlled.

Find out what's already
exploitable in your codebase.