Try each one before opening the answer. Every option is explained, with the CNCF documentation page that proves it.
Question 1 · choose 1
A penetration test reaches kube-scheduler's profiling endpoints from the node network. Following the Kubernetes hardening guide for the scheduler, what addresses the finding?
ABind kube-scheduler to localhost and set enableProfiling to false in its configuration
BPass --requestheader-client-ca-file so that requests must arrive through an authenticating proxy
CAdd the profiling paths to authorization-always-allow-paths so the scheduler authorizes them
DSet permit-port-sharing to true so that the scheduler shares its port with the API server
Show the answer and why
ABind kube-scheduler to localhost and set enableProfiling to false in its configuration
Correct
The guide says the scheduler rarely needs to be reachable from outside, so binding to localhost is secure, and that profiling should be disabled in production, through enableProfiling in the scheduler configuration.
BPass --requestheader-client-ca-file so that requests must arrive through an authenticating proxy
Incorrect
The guide says to avoid passing this argument to the scheduler at all.
CAdd the profiling paths to authorization-always-allow-paths so the scheduler authorizes them
Incorrect
Paths in that list are served to anonymous callers without an authorization check; adding the profiling endpoints there would open them further.
DSet permit-port-sharing to true so that the scheduler shares its port with the API server
Incorrect
The guide says to keep the default of false unless the security implications are understood; port sharing does not hide the profiling endpoints.
Profiling endpoints expose debugging information and can be abused for denial of service. The scheduler hardening guide recommends disabling profiling (enableProfiling: false, as the --profiling flag is deprecated) and binding the scheduler to localhost, since it seldom needs external access.
A custom scheduler profile enables a third-party plugin at the filter extension point. Why does the scheduler hardening guide ask administrators to review such plugins carefully?
AFilter plugins run with the kubelet's credentials on every node in the cluster
BFilter plugins can read the contents of Secrets in every namespace
CA filter plugin can mark every node unschedulable and halt new scheduling
DFilter plugins replace the scheduler's TLS configuration
Show the answer and why
AFilter plugins run with the kubelet's credentials on every node in the cluster
Incorrect
Scheduler plugins run inside the scheduler, not with kubelet credentials on nodes.
BFilter plugins can read the contents of Secrets in every namespace
Incorrect
The guide's concern for filter plugins is scheduling behavior, not Secret access.
CA filter plugin can mark every node unschedulable and halt new scheduling
Correct
The guide warns that plugins implementing prefilter or filter can mark all nodes as unschedulable, bringing scheduling of new Pods to a halt.
DFilter plugins replace the scheduler's TLS configuration
Incorrect
Plugins act on scheduling decisions; TLS settings come from the scheduler's command-line options.
Scheduler extension points such as queueSort, prefilter, filter and permit shape every scheduling decision. The hardening guide warns that a faulty filter plugin can make all nodes unschedulable, and suggests disabling those extension points when non-default plugins are used.
What does the Kubernetes scheduler hardening guide recommend for kube-scheduler's tls-cipher-suites option?
ALeave it empty so that the scheduler accepts every suite clients offer
BSet it only on the API server, since the scheduler serves no TLS
CCopy the cipher list from the etcd peer configuration
DAlways provide a list of preferred cipher suites
Show the answer and why
ALeave it empty so that the scheduler accepts every suite clients offer
Incorrect
Leaving the choice open is what the guide's advice is meant to prevent.
BSet it only on the API server, since the scheduler serves no TLS
Incorrect
The scheduler has its own secure serving options, and the guide covers its TLS settings.
CCopy the cipher list from the etcd peer configuration
Incorrect
Each component's TLS options are set on that component; the guide asks for a preferred list on the scheduler itself.
DAlways provide a list of preferred cipher suites
Correct
The guide says to always provide a list of preferred cipher suites, which ensures encryption never happens with insecure cipher suites.
Among the scheduler's TLS command-line options, the hardening guide asks administrators to always set tls-cipher-suites to a list of preferred suites, so that connections never fall back to insecure ones.