Google Cloud has addressed a medium-severity safety flaw in its platform that could possibly be abused by an attacker who already has entry to a Kubernetes cluster to escalate their privileges.
“An attacker who has compromised the Fluent Bit logging container might mix that entry with excessive privileges required by Anthos Service Mesh (on clusters which have enabled it) to escalate privileges within the cluster,” the corporate stated as a part of an advisory launched on December 14, 2023.
Palo Alto Networks Unit 42, which found and reported the shortcoming, stated adversaries might weaponize it to hold out “information theft, deploy malicious pods, and disrupt the cluster’s operations.”
From USER to ADMIN: Be taught How Hackers Acquire Full Management
Uncover the key techniques hackers use to turn into admins, find out how to detect and block it earlier than it is too late. Register for our webinar right now.
There isn’t a proof that the problem has been exploited within the wild. It has been addressed within the following variations of Google Kubernetes Engine (GKE) and Anthos Service Mesh (ASM) –
- 1.25.16-gke.1020000
- 1.26.10-gke.1235000
- 1.27.7-gke.1293000
- 1.28.4-gke.1083000
- 1.17.8-asm.8
- 1.18.6-asm.2
- 1.19.5-asm.4
A key prerequisite to efficiently exploiting the vulnerability hinges on an attacker having already compromised a FluentBit container by another preliminary entry strategies, equivalent to through a distant code execution flaw.
“GKE makes use of Fluent Bit to course of logs for workloads operating on clusters,” Google elaborated. “Fluent Bit on GKE was additionally configured to gather logs for Cloud Run workloads. The amount mount configured to gather these logs gave Fluent Bit entry to Kubernetes service account tokens for different Pods operating on the node.”
This meant {that a} risk actor might use this entry to achieve privileged entry to a Kubernetes cluster that has ASM enabled after which subsequently use ASM’s service account token to escalate their privileges by creating a brand new pod with cluster-admin privileges.
“The clusterrole-aggregation-controller (CRAC) service account might be the main candidate, as it could possibly add arbitrary permissions to current cluster roles,” safety researcher Shaul Ben Hai stated. “The attacker can replace the cluster position certain to CRAC to own all privileges.”
By the use of fixes, Google has eliminated Fluent Bit’s entry to the service account tokens and re-architected the performance of ASM to take away extreme role-based entry management (RBAC) permissions.
“Cloud distributors mechanically create system pods when your cluster is launched,” Ben Hai concluded. “They’re inbuilt your Kubernetes infrastructure, the identical as add-on pods which have been created whenever you allow a characteristic.”
“It’s because cloud or utility distributors usually create and handle them, and the person has no management over their configuration or permissions. This will also be extraordinarily dangerous since these pods run with elevated privileges.”
