CVE-2026-50564: Fission Environment CRD podspec passthrough enables hostPID/hostNetwork/privileged pods, node escape

Synopsis

This is a privilege-escalation and container-escape vulnerability in Fission, an open-source Kubernetes-native serverless framework. Any authenticated user with permission to create or update a Fission Environment custom resource can inject host-level settings (hostNetwork, hostPID, hostIPC, privileged containers, arbitrary serviceAccountName) into the pod spec that Fission builds and runs on the cluster. Successful exploitation gives the attacker a privileged pod with full visibility into host processes and networking, effectively escaping the container boundary and compromising the underlying Kubernetes node. HarborGuard tracks this advisory and will make a patched-image rebuild available the moment version 1.24.0 or a later fix is published upstream.

HarborGuard Coverage

Exploit Conditions

• Network reachabilityRequired

  The Fission API server is exposed over the network, so an attacker must be able to reach it remotely to submit a malicious Environment CRD.

• AuthenticationRequired

  The attacker must hold a low-privilege Kubernetes account with permission to create or update Fission Environment resources; anonymous access is not sufficient.

• Victim interactionNot required

  No user interaction is needed; the attacker submits the malicious resource directly and the framework processes it automatically.

• Attack complexityDetail

  Attack complexity is low: the exploit requires only a well-formed Environment CRD payload and no race conditions, memory layout assumptions, or other environmental factors.

Blast Radius

• Attacker gains a privileged container with hostPID enabled, allowing it to inspect and signal every process running on the underlying Kubernetes node.
• With hostNetwork enabled, the attacker's pod shares the node's network stack, exposing all inter-node traffic, cloud metadata endpoints, and internal cluster services.
• Arbitrary serviceAccountName injection lets the attacker impersonate any Kubernetes service account on the cluster, reading secrets and calling the API server with that account's permissions.
• Full node compromise allows the attacker to read host filesystem paths, extract credentials stored in other pods, and potentially pivot to additional nodes in the cluster.