Try each one before opening the answer. Every option is explained, with the CNCF documentation page that proves it.
Question 1 · choose 1
Several teams share the worker nodes of one cluster. Team A pulls its private image registry.example.com/a/billing:3.2 with an imagePullSecret. The kubelets are configured with imagePullCredentialsVerificationPolicy: NeverVerify, the setting of the beta image pull credential verification feature that skips the check for images already on a node, and the node team will not change kubelet configuration. Team B has no credentials for that registry, yet a Pod of Team B that names the same image without a pull secret starts whenever it lands on a node where Team A's Pod ran before. The fix must not depend on Team B's cooperation. What closes the gap?
AEnforce the Restricted Pod Security Standard in Team B's namespace
BEnable the AlwaysPullImages admission controller on the API server
CSet imagePullPolicy: Always on every container of Team A's Pods
DMake Team A's pull Secret readable only by Team A's service account
Show the answer and why
AEnforce the Restricted Pod Security Standard in Team B's namespace
Incorrect
Pod Security Standards check security-related Pod fields such as capabilities and privilege; they do not check whether a Pod is entitled to an image.
BEnable the AlwaysPullImages admission controller on the API server
Correct
AlwaysPullImages modifies every new Pod to force the pull policy to Always, so images are pulled before containers start and valid credentials are required each time, whoever creates the Pod.
CSet imagePullPolicy: Always on every container of Team A's Pods
Incorrect
That makes Team A's own starts pull from the registry, but the pull policy belongs to each Pod: Team B's Pods keep their own policy and still start from the cached image.
DMake Team A's pull Secret readable only by Team A's service account
Incorrect
The credential was never the problem: once an image is on a node, any Pod can use it by name without any authorization check, whoever can read the Secret.
The admission controller reference describes this gap: without AlwaysPullImages, once an image has been pulled to a node any Pod from any user can use it by knowing its name. Since Kubernetes v1.35 the kubelet re-verifies credentials by default, but with the NeverVerify policy that check is off, so forcing the Always pull policy on every new Pod at admission is what makes each start authenticate to the registry.
Developers pull base images straight from public registries into their builds. What registry setup does the CNCF Cloud Native Security Whitepaper recommend instead?
AOne public registry for everything, with tags locked after the first push of each image
BPulling public images directly on each node to avoid an extra copy
CA registry per node, filled from public registries as Pods start
DStaged registries: authorized developers import public base images internally
Show the answer and why
AOne public registry for everything, with tags locked after the first push of each image
Incorrect
Locking tags does not control what enters from public sources, which the staged setup is meant to do.
BPulling public images directly on each node to avoid an extra copy
Incorrect
Direct pulls from public sources are exactly what the staged setup puts under control.
CA registry per node, filled from public registries as Pods start
Incorrect
Per-node caches of public images do not add the provenance control the whitepaper describes.
DStaged registries: authorized developers import public base images internally
Correct
The whitepaper recommends several registry stages: authorized developers pull public base images into an internal registry, with team registries and a pre-production registry for images ready for production.
Because open source components often come from public sources, the whitepaper advises staged registries: controlled import of public base images into an internal registry, per-team development registries and a staging registry for production-ready images, with access control on each.
What does the Kubernetes security checklist say about shells and debugging utilities in production container images?
ALeave them out, and use an ephemeral debug container for troubleshooting
BInclude them, so that kubectl exec works during incidents
CInclude them only in images that run as root
DKeep them, but scan the image more often
Show the answer and why
ALeave them out, and use an ephemeral debug container for troubleshooting
Correct
The checklist says production images should not contain shells or debugging utilities, since an ephemeral debug container can be used for troubleshooting.
BInclude them, so that kubectl exec works during incidents
Incorrect
The checklist recommends minimal images and points to ephemeral containers instead of tools baked into the image.
CInclude them only in images that run as root
Incorrect
Running as root makes extra tools more dangerous, not less; the checklist leaves them out of production images.
DKeep them, but scan the image more often
Incorrect
Scanning does not remove the extra attack surface that unused tools add.
Images should contain the bare minimum to run the program. The checklist advises building from a minimal base and leaving out shells and debugging tools in production, relying on ephemeral debug containers when troubleshooting is needed.