Skip to content

KCSA · Domain 5: Platform Security · 16% of the exam

Task 5.3: Observability

Security monitoring: audit logs, events, metrics, runtime detection and alerting on suspicious cluster behaviour.

Study it

Sample questions

Try each one before opening the answer. Every option is explained, with the CNCF documentation page that proves it.

Question 1 · choose 1

A security operations team feeds cluster audit events and workload telemetry into an observability platform with a machine learning anomaly detector. Analysts are overwhelmed by alerts, and many of them concern rare administrative actions that were planned in advance through change tickets. Which way of handling anomalies follows the CNCF Cloud Native Security Whitepaper?

  1. ALet the detector close alerts it classes as expected and drop those events from the records
  2. BTurn off automated detection and send every event to analysts, who can reason about context
  3. CExclude planned administrative actions from detection, since change tickets already approve them
  4. DLabel expected anomalies, tag unexpected ones for review, keep humans involved
Show the answer and why
  • ALet the detector close alerts it classes as expected and drop those events from the records

    Incorrect

    The whitepaper says expected anomalies should be tracked and labeled for auditing purposes; dropping them removes that record.

  • BTurn off automated detection and send every event to analysts, who can reason about context

    Incorrect

    The whitepaper says rule engines and AI/ML may automate some anomaly detection for scalability; only the replacement of humans is ruled out.

  • CExclude planned administrative actions from detection, since change tickets already approve them

    Incorrect

    The whitepaper names administrative actions that are rarely performed among the anomalies that must be identified and tracked.

  • DLabel expected anomalies, tag unexpected ones for review, keep humans involved

    Correct

    The whitepaper says expected anomalies should be tracked and labeled for auditing, unexpected ones tagged for additional review, and that automated detection should not yet replace humans.

The whitepaper's guidance on anomalies is to identify and track them, including rarely performed administrative actions, label the expected ones for auditing and tag the unexpected ones for further review. Rule engines and AI/ML can help detection scale, but should not yet replace people who can reason about complex cases.

Question 2 · choose 1

How does the CNCF Cloud Native Security Whitepaper suggest teams use observability data to find security problems?

  1. ABy deleting metrics older than a day to reduce noise
  2. BBehavioral and heuristic analysis to detect outliers
  3. CBy disabling logs for system components
  4. DBy collecting metrics only during business hours
Show the answer and why
  • ABy deleting metrics older than a day to reduce noise

    Incorrect

    Discarding data does not help detect anomalies.

  • BBehavioral and heuristic analysis to detect outliers

    Correct

    The whitepaper says behavioral and heuristic analysis lets teams detect and escalate outliers, suspicious events and unexplained calls.

  • CBy disabling logs for system components

    Incorrect

    System component logs are a source of security insight, not noise to remove.

  • DBy collecting metrics only during business hours

    Incorrect

    Attacks are not limited to business hours; gaps would hide them.

Observability gives security teams the data to spot anomalies. The whitepaper encourages behavioral and heuristic analysis, possibly helped by statistical or machine learning models, to surface outliers for the right people.

Question 3 · choose 1

Which item about audit logs does the Kubernetes security checklist include under Logs and auditing?

  1. AAudit logs are disabled to save disk space
  2. BAudit logs are readable by every namespace owner
  3. CAudit logs, if enabled, are protected from general access
  4. DAudit logs are kept inside the Pods that generated the requests
Show the answer and why
  • AAudit logs are disabled to save disk space

    Incorrect

    The checklist does not suggest disabling audit logs.

  • BAudit logs are readable by every namespace owner

    Incorrect

    The checklist asks for them to be protected from general access.

  • CAudit logs, if enabled, are protected from general access

    Correct

    The checklist's Logs and auditing item reads that audit logs, if enabled, are protected from general access.

  • DAudit logs are kept inside the Pods that generated the requests

    Incorrect

    Audit logs are written by the API server, not by Pods.

Audit logs contain detailed records of cluster activity and can contain sensitive data. The security checklist asks that, where enabled, they are protected from general access.

Practice domain 5 →Practice all domains →