A flaw was found in Red Hat Openshift AI Service. A low-privileged attacker with access to an authenticated account, for example as a data scientist using a standard Jupyter notebook, can escalate their privileges to a full cluster administrator. This allows for the complete compromise of the cluster's confidentiality, integrity, and availability. The attacker can steal sensitive data, disrupt all services, and take control of the underlying infrastructure, leading to a total breach of the platform and all applications hosted on it.
RHOAI versions 2.16.x
The recommended fix is to upgrade to version 2.16.3.
However, if Kueue features are not required, the Kueue component management state can be set to "Removed" in the RHOAI DataScienceCluster resource:
~~~
oc patch dsc default-dsc --type=merge -p='{"spec":{"components":{"kueue":{"managementState":"Removed"}}}}' -n redhat-ods-operator
~~~
RHOAI versions 2.19+
The recommended fix is to apply the available erratum.
Alternatively, follow these steps:
1. Prevent the RHOAI operator from managing the kueue-batch-user-rolebinding by applying the necessary annotation:
~~~
oc patch clusterrolebinding kueue-batch-user-rolebinding -p '{"metadata":{"annotations": {"opendatahub.io/managed":"false"}}}'
~~~
- Disable the ClusterRoleBinding by updating its subject to a different, non-existent, group:
~~~
oc patch clusterrolebinding kueue-batch-user-rolebinding \
-p '{
"subjects": [
{
"kind": "Group",
"name": "REPLACEME",
"apiGroup": "rbac.authorization.k8s.io"
}
]
}'
~~~
It is important that the group used for the subject does not exist on the cluster to prevent the risk of unintentionally assigning these permissions to other non-privileged users.
- Once updates providing fixes have been applied, it's recommended to remove the clusterrolebinding created in step 2 of the mitigation.
~~~
oc delete clusterrolebinding kueue-batch-user-rolebinding
~~~