CareerExamPro

Certified Kubernetes Administrator & Developer (CKA/CKAD) practice test

10 practice questions for this exam, each with a full explanation and cited rule or blueprint reference.

Start practising

ServiceNow, Databricks, Snowflake & K8s: practice drill

10 realistic exam questions, graded as you go, with the exact rule or statute citation for every answer. Start free, no sign-up.

Timed mock exam

10 questions in 12 minutes, drawn from every topic. No answers shown until the end, then a full score breakdown, your weakest topics and a review of what you missed.

Practice tests that prepare you for this exam

Practise by topic

About this exam

Agency / Issuer
CNCF / Linux Foundation
Region
Global / Online

Exact question counts, passing scores, and allowed reference materials vary by testing window and state/agency. Always confirm current rules with the official testing provider.

Sample questions

SQL
ALTER WAREHOUSE bi_wh SET
  MIN_CLUSTER_COUNT = 1
  MAX_CLUSTER_COUNT = 4
  SCALING_POLICY = 'STANDARD';

A BI team runs 120 concurrent dashboard queries every Monday at 9:00 AM against a MEDIUM Snowflake Virtual Warehouse. Query Profile shows fast execution once queries start, but heavy Queued (Overload) wait times. Which change directly addresses this bottleneck?

  1. Resize bi_wh up to 2X-LARGE with a single cluster, because vertical scaling is the only way Snowflake increases query concurrency slots.
  2. Scale out with a multi-cluster warehouse as shown, adding clusters automatically to absorb concurrent queuing.
  3. Enable AUTO_SUSPEND = 60 without multi-cluster scaling, because suspending idle clusters clears the concurrency queue faster.
  4. Switch SCALING_POLICY to ECONOMY on a single-cluster warehouse so Snowflake automatically splits large joins across regions.
Show the answer

B. Scale out with a multi-cluster warehouse as shown, adding clusters automatically to absorb concurrent queuing.

Scaling out (multi-cluster warehouse) adds identical clusters to handle high concurrency and eliminate Queued (Overload) delays. Scaling up (larger t-shirt size) increases memory and CPU per cluster to accelerate individual complex queries or reduce spilling, not primarily concurrency queuing.

Reference: Snowflake Virtual Warehouses: Scaling Up vs. Scaling Out

An organization implements Snowflake RBAC with a custom DATA_ENG_ROLE that creates databases and tables:

SQL
USE ROLE USERADMIN;
CREATE ROLE data_eng_role;
USE ROLE SECURITYADMIN;
GRANT ROLE data_eng_role TO ROLE SYSADMIN;

Why does Snowflake's RBAC best practice explicitly grant custom roles to SYSADMIN rather than leaving them unattached or creating objects as ACCOUNTADMIN?

  1. SYSADMIN is the only system role permitted to run GRANT USAGE or MANAGE GRANTS across databases and schemas.
  2. Custom roles cannot own tables or warehouses in Snowflake unless they directly inherit ACCOUNTADMIN privileges.
  3. Leaving a custom role unattached automatically grants its object ownership to PUBLIC, exposing tables to all account users.
  4. It lets SYSADMIN manage all custom-role objects while keeping ACCOUNTADMIN for account-level governance.
Show the answer

D. It lets SYSADMIN manage all custom-role objects while keeping ACCOUNTADMIN for account-level governance.

In Snowflake RBAC, SYSADMIN should sit at the top of the custom role hierarchy so admins can manage all objects owned by custom roles without needing ACCOUNTADMIN. SECURITYADMIN manages grants (MANAGE GRANTS), while ACCOUNTADMIN is reserved for billing, security policies, and account-level settings.

Reference: Snowflake Access Control (RBAC): System-Defined Role Hierarchy

A data processing worker Pod repeatedly restarts. Running kubectl describe pod spark-worker-0 displays:

YAML
State:          Waiting
  Reason:       CrashLoopBackOff
Last State:     Terminated
  Reason:       OOMKilled
  Exit Code:    137
Limits:
  memory:       512Mi

What is the root cause and primary remediation?

  1. The container image failed registry authentication (ImagePullBackOff); create a docker-registry Secret and attach it via imagePullSecrets.
  2. The kubelet liveness probe timed out on TCP port 8080; increase initialDelaySeconds to prevent exit code 137.
  3. The container exceeded its 512Mi memory limit and received SIGKILL from the kernel OOM killer; raise limits.memory or lower app memory usage.
  4. The Pod requested more CPU cores than the node has allocatable; lower resources.requests.cpu so the scheduler stops evicting it.
Show the answer

C. The container exceeded its 512Mi memory limit and received SIGKILL from the kernel OOM killer; raise limits.memory or lower app memory usage.

Exit code 137 (128 + 9, SIGKILL) paired with Reason: OOMKilled confirms the Linux kernel OOM killer terminated the container for exceeding its 512Mi cgroup memory limit, triggering CrashLoopBackOff. Raising resources.limits.memory and tuning the worker's heap/memory footprint resolves it.

Reference: Kubernetes Troubleshooting: OOMKilled (Exit 137) vs. CrashLoopBackOff

Taking this exam soon?

Tell us which exam domains or calculation drills you want more practice on.

Other Enterprise Cloud & Data exams