Practice for the Red Hat Certified Specialist in OpenShift AI (EX267) exam with 17 exam-style practice questions, instant answer reveals, and concise explanations of every correct answer. Topics include: What configuration error causes an external inference request to return a 503 Service Unavailable code?. Follow @CertPunch and visit certpunch.com for more certification practice exams and study content.
Prefer hands-on? Take this round as an interactive practice test — answer every question, get instant feedback, and see your score: Start the Red Hat Certified Specialist in OpenShift AI (EX267) practice test →
What you will practice
- What configuration error causes an external inference request to return a 503 Service Unavailable code?
- What type of validation is performed by the OpenShift AI pipeline compiler before a pipeline yaml is exported?
- What is the default resource configuration strategy for OpenShift AI workbenches regarding requests and limit…
- Which metric tracks the shifting distribution of independent variables in production data compared to trainin…
- What configuration in the DataScienceCluster CR acts as the master toggle to remove a component cleanly?
- Which CR property allows cluster admins to selectively toggle components (e.g. TrustyAI) as managed or remove…
Answers and explanations
Tap a question to expand the answer and the exam reasoning. Try to commit to your own pick first.
Q1. What configuration error causes an external inference request to return a 503 Service Unavailable code?
Answer: C. The underlying model serving runtime pods are failing or not fully initialized
A 503 Service Unavailable code occurs when the underlying model serving runtime pods are failing or not fully initialized. Distractors like expired tokens or incorrect filenames usually cause build or access errors rather than routing failures.
Q2. What type of validation is performed by the OpenShift AI pipeline compiler before a pipeline yaml is exported?
Answer: D. Checking structural syntax correctness and dependency graph connectivity
The OpenShift AI pipeline compiler checks structural syntax correctness and dependency graph connectivity before exporting the YAML. It cannot validate external cluster states like GPU availability or S3 credentials at compile time.
Q3. What is the default resource configuration strategy for OpenShift AI workbenches regarding requests and limits?
Answer: D. Defining specific minimum requests and maximum limits for CPU and memory stability
OpenShift AI workbenches use specific minimum resource requests and maximum limits for CPU and memory to ensure cluster stability. Configuring infinite boundaries or removing limits risks node instability and scheduling failures.
Q4. Which metric tracks the shifting distribution of independent variables in production data compared to training baseline datasets?
Answer: C. Covariate Data Drift metrics evaluation values
Covariate data drift metrics evaluate shifts in the distribution of independent variables between production and training data. F1 accuracy measures model performance, while disparate impact evaluates fairness across specific groups.
Q5. What configuration in the DataScienceCluster CR acts as the master toggle to remove a component cleanly?
Answer: C. managementState: Removed
Setting managementState to Removed cleanly uninstalls a specific component in the DataScienceCluster custom resource. This is the standard operator lifecycle practice for disabling modules without causing deployment errors.
Q6. Which CR property allows cluster admins to selectively toggle components (e.g. TrustyAI) as managed or removed?
Answer: D. DataScienceCluster components section spec
The spec.components section of the DataScienceCluster CR allows admins to toggle individual components like TrustyAI. Remember that DSCInitialization handles cluster-wide prerequisites, while DataScienceCluster manages the individual features.
Q7. Which environment variable determines the configuration directory location for Jupyter runtimes?
Answer: A. JUPYTER_CONFIG_DIR
The JUPYTER_CONFIG_DIR environment variable specifies the directory path for Jupyter configuration files. This is essential when building custom workbench images that require specific runtime defaults or overrides.
Q8. What system user group assignment gives members comprehensive overview control across the OpenShift AI platform dashboard configs?
Answer: A. ods-admins
The ods-admins group grants comprehensive administrative control over the OpenShift AI dashboard and platform configurations. Assigning users here is the standard way to delegate cluster-level administration without making them full cluster operators.
Q9. Which log stream should be analyzed to troubleshoot authentication validation failures between OpenShift AI and an external OAuth provider?
Answer: D. OAuth operator pods or cluster authentication subsystem logs
Authentication validation failures between OpenShift AI and an external OAuth provider must be analyzed using the OAuth operator or cluster authentication subsystem logs. Workbench or pipeline logs will not show cluster-level identity provider integration errors.
Q10. What configuration option allows an administrator to prevent users from modifying global environment variable layouts?
Answer: D. Setting predefined variables inside the cluster-wide OdhDashboardConfig properties
Configuring the OdhDashboardConfig custom resource locks down predefined cluster settings. Standard namespace policies or terminal deletion cannot prevent users from overriding global dashboard environment configurations.
Q11. What must be modified to enable a graphical hardware selection type dropdown in the dashboard?
Answer: B. AcceleratorProfile CRD
Creating an AcceleratorProfile custom resource exposes hardware options in the dashboard interface. DashboardConfig maps handle general feature toggles, but accelerator profiles specifically define the available GPU selections.
Q12. Which microservice tracks the end-to-end execution paths of an inference request across a complex multi-model mesh chain?
Answer: C. OpenShift Service Mesh distributed tracing components (Jaeger / Tempo)
OpenShift Service Mesh leverages distributed tracing tools like Jaeger or Tempo to track inference requests across the mesh. Elyra only handles pipeline orchestration and cannot monitor live multi-model mesh traffic or execution paths.
Q13. WHICH TOOL DIRECTLY CONVERTS INTERACTIVE IPYNB NOTEBOOKS INTO COMPILED PIPELINE STEPS NATIVELY IN RHOAI?
Answer: B. Elyra Extension tools suite
The Elyra extension directly converts interactive IPYNB notebooks into compiled pipeline steps within the workbench. S2I only builds images from raw source code, not complex pipeline workflows.
Q14. Which underlying engine coordinates the rolling update strategy when a model file is updated in object storage?
Answer: B. ModelMesh Controller or KServe Controller depending on configuration
Depending on the deployment configuration, either the ModelMesh or KServe controller coordinates rolling updates when storage changes. Tekton only handles pipeline execution and does not manage serving runtime deployments.
Q15. Which parameter in the KServe configuration defines the fallback container image used when a custom runtime image is missing?
Answer: A. The default images defined in the KServe controller configuration ConfigMap/CR
KServe uses default images defined in its controller configuration to pull runtime containers when a custom image is missing. Relying on a random host image or registry stream is not how OpenShift AI manages its serving runtimes.
Q16. Which property in AcceleratorProfiles matches workloads to the nodes containing corresponding hardware chips?
Answer: A. spec.nodeSelector block fields defining explicit node match criteria attributes
The spec.nodeSelector block in an AcceleratorProfile matches workloads to nodes with specific hardware chips. Tolerations only allow pods to bypass taints, while node selectors actively target the required accelerator hardware.
Q17. Which parameters are required inside a pipeline definition to configure an automated cron-based recurring execution schedule?
Answer: D. The cron expression string and max concurrent runs parameters defined in the pipeline run trigger menu
Triggering a pipeline via cron requires a cron expression string and a max concurrent runs parameter to manage execution schedules. Authentication tokens and memory sizing are unrelated to scheduling recurring runs.
More Red Hat Certified Specialist in OpenShift AI (EX267) drills and other practice exams are on @CertPunch. New rounds drop every few days at certpunch.com.