CVE-2023-22649: Rancher 'Audit Log' leaks sensitive information

A vulnerability has been identified which may lead to sensitive data being leaked into Rancher’s audit logs. Rancher Audit Logging is an opt-in feature, only deployments that have it enabled and have AUDIT_LEVEL set to 1 or above are impacted by this issue.

The leaks might be caught in the audit logs upon these actions:

• Creating cloud credentials or new authentication providers. It is crucial to note that all authentication providers (such as AzureAD) and cloud providers (such as Google) are impacted.
• Downloading a kubeconfig file from a downstream or a local cluster.
• Logging in/out from Rancher.

The affected data may include the following:

• HTTP headers

• API Server calls returning Secret objects (including sub-types, such as kubernetes.io/dockerconfigjson).
• Raw command lines used by agents to connect to the Rancher server which expose sensitive information (e.g. register ... --token abc).
• Kubeconfig contents when the ‘Download KubeConfig’ feature is used in the Rancher UI.

The patched versions will redact the sensitive data, replacing it with [redacted], making it safer for consumption. It is recommended that static secrets are rotated after the system is patched, to limit the potential impact of sensitive data being misused due to this vulnerability.

Note:

1. The severity of the vulnerability is intricately tied to the logging strategy employed. If logs are kept locally (default configuration), the impact is contained within the system, limiting the exposure. However, when logs are shipped to an external endpoint, the vulnerability’s severity might increase, as resistance against leaks is contingent on the security measures implemented at the external log collector level.
2. The final impact severity for confidentiality, integrity and availability is dependent on the permissions that the leaked credentials have on their own services.