Description

IBM Observability with Instana (Agent) Build 1.0.303 through 1.0.323 IBM Instana Agent Operator could allow an authenticated remote attacker to obtain sensitive information, caused by missing destination namespace validation when copying etcd mTLS client credentials from the openshift-etcd system namespace into an attacker-controlled namespace.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "7.7",
  "pubDate": "2026-09-04T16:17:21.777Z",
  "pubdate": "2026-09-04T16:17:21.777Z",
  "executiveSummary": "IBM Observability with Instana (Agent) Build 1.0.303 through 1.0.323 contains an information disclosure vulnerability in the IBM Instana Agent Operator. An authenticated remote attacker can exploit this vulnerability to obtain highly sensitive etcd mutual TLS (mTLS) client credentials. The security flaw is caused by a missing destination namespace validation check when the operator copies credentials from the secure 'openshift-etcd' system namespace into other namespaces.\nThis vulnerability poses a severe security risk in multi-tenant Kubernetes and OpenShift environments. By manipulating the operator, an attacker with authenticated access can force the replication of cryptographic secrets to a namespace under their direct control. Successful exploitation grants the attacker the credentials required to access the central etcd database directly, bypassing standard cluster access controls and potentially leading to a complete compromise of the container orchestration platform.",
  "technicalDetails": "The vulnerability is located within the credential replication logic of the IBM Instana Agent Operator, which is active in IBM Observability with Instana (Agent) Build 1.0.303 through 1.0.323. Under normal conditions, the operator manages the deployment and configuration of Instana agents, which may require accessing metrics or configuration data from system namespaces, including 'openshift-etcd'. Because the operator runs with elevated cluster-wide privileges, it has the authorized capability to read secrets from system namespaces and write them to target agent namespaces.\nHowever, the operator fails to validate the destination namespace when executing these copy operations. The reconciliation loop of the operator trusts user-supplied or Custom Resource (CR) configurations without verifying if the target namespace is authorized to receive administrative system credentials. An authenticated attacker can exploit this design flaw by deploying or modifying an InstanaAgent CR within a namespace they control, specifying their local namespace as the target for the agent deployment.\nWhen the operator reconciles the state, it automatically copies the etcd mTLS client credentials—consisting of sensitive client certificates and private keys—from the 'openshift-etcd' namespace into the attacker's namespace. Once the secret is written to the destination, the attacker can extract the private key and certificate using standard API queries.\nWith the extracted mTLS credentials, the attacker can establish a direct connection to the etcd key-value store, completely bypassing the Kubernetes API server and its Role-Based Access Control (RBAC) mechanisms. Since etcd is the primary data store for the cluster containing all configuration data, secrets, and service account tokens, possessing direct read and write access to etcd allows the attacker to retrieve all sensitive data, manipulate cluster state, and achieve full cluster-administrator privileges."
}