Kubernetes node setting allows for neighboring hosts to bypass localhost boundary
The Kubelet and kube-proxy components in versions 1.1.0-1.16.10, 1.17.0-1.17.6, and 1.18.0-1.18.3 were found to contain a security issue which allows adjacent hosts to reach TCP and UDP services bound to 127.0.0.1 running on the node or in the node's network namespace. Such a service is generally thought to be reachable only by other processes on the same host, but due to this defeect, could be reachable by other hosts on the same LAN as the node, or by containers running on the same node as the service.
Kubernetes node disk Denial of Service by writing to container /etc/hosts
The Kubernetes kubelet component in versions 1.1-1.16.12, 1.17.0-1.17.8 and 1.18.0-1.18.5 do not account for disk usage by a pod which writes to its own /etc/hosts file. The /etc/hosts file mounted in a pod by kubelet is not included by the kubelet eviction manager when calculating ephemeral storage usage by a pod. If a pod writes a large amount of data to the /etc/hosts file, it could fill the storage space of the node and cause the node to fail.
The Kubernetes kube-controller-manager in versions v1.0-v1.17 is vulnerable to a credential leakage via error messages in mount failure logs and events for AzureFile and CephFS volumes.
Privilege escalation from compromised node to cluster
The Kubernetes kube-apiserver in versions v1.6-v1.15, and versions prior to v1.16.13, v1.17.9 and v1.18.6 are vulnerable to an unvalidated redirect on proxied upgrade requests that could allow an attacker to escalate privileges from a node compromise to a full cluster compromise.
The Kubernetes kube-controller-manager in versions v1.0-1.14, versions prior to v1.15.12, v1.16.9, v1.17.5, and version v1.18.0 are vulnerable to a Server Side Request Forgery (SSRF) that allows certain authorized users to leak up to 500 bytes of arbitrary information from unprotected endpoints within the master's host network (such as link-local or loopback services).
Kubernetes API Server denial of service vulnerability from malicious YAML payloads
The Kubernetes API Server component in versions 1.1-1.14, and versions prior to 1.15.10, 1.16.7 and 1.17.3 allows an authorized user who sends malicious YAML payloads to cause the kube-apiserver to consume excessive CPU cycles while parsing YAML.
The Kubernetes API server component in versions prior to 1.15.9, 1.16.0-1.16.6, and 1.17.0-1.17.2 has been found to be vulnerable to a denial of service attack via successful API requests.
The Kubelet component in versions 1.15.0-1.15.9, 1.16.0-1.16.6, and 1.17.0-1.17.2 has been found to be vulnerable to a denial of service attack via the kubelet API, including the unauthenticated HTTP read-only API typically served on port 10255, and the authenticated HTTPS API typically served on port 10250.
The Kubernetes kubectl cp command in versions 1.1-1.12, and versions prior to 1.13.11, 1.14.7, and 1.15.4 allows a combination of two symlinks provided by tar output of a malicious container to place a file outside of the destination directory specified in the kubectl cp invocation. This could be used to allow an attacker to place a nefarious file using a symlink, outside of the destination tree.
Kubernetes API server follows unvalidated redirects from streaming Kubelet endpoints
Improper validation of URL redirection in the Kubernetes API server in versions prior to v1.14.0 allows an attacker-controlled Kubelet to redirect API server requests from streaming endpoints to arbitrary hosts. Impacted API servers will follow the redirect as a GET request with client-certificate credentials for authenticating to the Kubelet.
Docker Engine before 1.8.3 and CS Docker Engine before 1.6.2-CS7 does not properly validate and extract the manifest object from its JSON representation during a pull, which allows attackers to inject new attributes in a JSON object and bypass pull-by-digest validation.
Docker Engine before 1.8.3 and CS Docker Engine before 1.6.2-CS7 do not use a globally unique identifier to store image layers, which makes it easier for attackers to poison the image cache via a crafted image in pull or push commands.
Kubernetes API Server JSON/YAML parsing vulnerable to resource exhaustion attack
Improper input validation in the Kubernetes API server in versions v1.0-1.12 and versions prior to v1.13.12, v1.14.8, v1.15.5, and v1.16.2 allows authorized users to send malicious YAML or JSON payloads, causing the API server to consume excessive CPU or memory, potentially crashing and becoming unavailable. Prior to v1.14.0, default RBAC policy authorized anonymous users to submit requests that could trigger this vulnerability. Clusters upgraded from a version prior to v1.14.0 keep the more permissive policy by default for backwards compatibility.
Kubernetes client-go logs authorization headers at debug verbosity levels
The Kubernetes client-go library logs request headers at verbosity levels of 7 or higher. This can disclose credentials to unauthorized users via logs or command output. Kubernetes components (such as kube-apiserver) prior to v1.16.0, which make use of basic or bearer token authentication, and run at high verbosity levels, are affected.
The kubectl cp command allows copying files between containers and the user machine. To copy files from a container, Kubernetes runs tar inside the container to create a tar archive, copies it over the network, and kubectl unpacks it on the user’s machine. If the tar binary in the container is malicious, it could run any code and output unexpected, malicious results. An attacker could use this to write files to any path on the user’s machine when kubectl cp is called, limited only by the system permissions of the local user. Kubernetes affected versions include versions prior to 1.13.9, versions prior to 1.14.5, versions prior to 1.15.2, and
Kubernetes kubelet exposes /debug/pprof info on healthz port
The debugging endpoint /debug/pprof is exposed over the unauthenticated Kubelet healthz port. The go pprof endpoint is exposed over the Kubelet's healthz port. This debugging endpoint can potentially leak sensitive information such as internal Kubelet memory addresses and configuration, or for limited denial of service. Versions prior to 1.15.0, 1.14.4, 1.13.8, and 1.12.10 are affected. The issue is of medium severity, but not exposed by the default configuration.
Kubernetes kube-apiserver allows access to custom resources via wrong scope
The Kubernetes kube-apiserver mistakenly allows access to a cluster-scoped custom resource if the request is made as if the resource were namespaced. Authorizations for the resource accessed in this manner are enforced using roles and role bindings within the namespace, meaning that a user with access only to a resource in one namespace could create, view update or delete the cluster-scoped resource (according to their namespace role privileges). Kubernetes affected versions include versions prior to 1.13.9, versions prior to 1.14.5, versions prior to 1.15.2, and versions 1.7, 1.8, 1.9, 1.10, 1.11, 1.12.
The kubectl cp command allows copying files between containers and the user machine. To copy files from a container, Kubernetes runs tar inside the container to create a tar archive, copies it over the network, and kubectl unpacks it on the user’s machine. If the tar binary in the container is malicious, it could run any code and output unexpected, malicious results. An attacker could use this to write files to any path on the user’s machine when kubectl cp is called, limited only by the system permissions of the local user. Kubernetes affected versions include versions prior to 1.12.9, versions prior to 1.13.6, versions prior to 1.14.2, and
kubelet-started container uid changes to root after first restart or if image is already pulled to the node
In kubelet v1.13.6 and v1.14.2, containers for pods that do not specify an explicit runAsUser attempt to run as uid 0 (root) on container restart, or if the image was previously pulled to the node. If the pod specified mustRunAsNonRoot: true, the kubelet will refuse to start the container as root. If the pod did not specify mustRunAsNonRoot: true, the kubelet will run the container as uid 0.
In Kubernetes v1.8.x-v1.14.x, schema info is cached by kubectl in the location specified by --cache-dir (defaulting to $HOME/.kube/http-cache), written with world-writeable permissions (rw-rw-rw-). If --cache-dir is specified and pointed at a different location accessible to other users/groups, the written files may be modified by other users/groups and disrupt the kubectl invocation.