Try each one before opening the answer. Every option is explained, with the CNCF documentation page that proves it.
Question 1 · choose 1
Auditors of a self-managed control plane find that kube-controller-manager authenticates as a user bound to cluster-admin and runs every control loop under that single identity. They want each built-in controller to call the API with only the permissions of its own role, so that a flaw in one loop cannot use the rights of the others. What achieves this?
ABind the controller manager's user to each system:controller: role one by one
BRun several controller manager processes, each with --controllers naming one controller
CStart kube-controller-manager with --use-service-account-credentials
DReplace the cluster-admin binding with the default system:kube-controller-manager role
Show the answer and why
ABind the controller manager's user to each system:controller: role one by one
Incorrect
The roles would still all apply to one credential, so every loop keeps the union of their permissions.
BRun several controller manager processes, each with --controllers naming one controller
Incorrect
Splitting the processes does not split identities: each process would still authenticate with the same controller manager credential and its combined permissions.
CStart kube-controller-manager with --use-service-account-credentials
Correct
With this flag the controller manager starts each controller using a separate service account, and matching system:controller: roles exist for the built-in controllers.
DReplace the cluster-admin binding with the default system:kube-controller-manager role
Incorrect
Dropping cluster-admin narrows the rights, but the controller manager role covers what the whole component needs; one identity still holds the permissions of every loop.
Without --use-service-account-credentials, kube-controller-manager runs all control loops with its own credential, which must hold every relevant role. With it, each controller uses its own service account bound to its system:controller: role, and the security checklist lists the flag as a baseline item. Narrowing the shared credential's role still leaves one identity for all loops.
Which control plane component's --cluster-signing-cert-file flag names the CA certificate used to issue cluster-scoped certificates?
Akube-apiserver
Bkube-scheduler
Ckubelet
Dkube-controller-manager
Show the answer and why
Akube-apiserver
Incorrect
The API server's TLS flags configure its own serving certificate; it does not sign certificates for CSRs.
Bkube-scheduler
Incorrect
The scheduler does not sign certificates.
Ckubelet
Incorrect
The kubelet requests certificates; it does not issue them for the cluster.
Dkube-controller-manager
Correct
The kube-controller-manager reference describes --cluster-signing-cert-file as the CA certificate used to issue cluster-scoped certificates.
Signing of approved CertificateSigningRequests for the built-in signers is done by the controller manager, which uses the CA given in --cluster-signing-cert-file and its key.
Which controller ensures that a ServiceAccount named default exists in every active namespace?
AThe Namespace controller
BThe Node controller
CThe token cleaner
DThe ServiceAccount controller
Show the answer and why
AThe Namespace controller
Incorrect
The Namespace controller handles namespace lifecycle such as deletion, not the default ServiceAccount.
BThe Node controller
Incorrect
The Node controller manages Node objects and their health.
CThe token cleaner
Incorrect
The token cleaner works on bootstrap tokens, not on ServiceAccounts.
DThe ServiceAccount controller
Correct
The documentation says the ServiceAccount controller manages ServiceAccounts inside namespaces and ensures one named default exists in every active namespace.
Part of kube-controller-manager, the ServiceAccount controller makes sure every active namespace has a default ServiceAccount, which Pods use when they name no other.