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: Which components are deployed by default when the DataScienceCluster resource sets the 'pipelines' component state to 'M. 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
- Which components are deployed by default when the DataScienceCluster resource sets the 'pipelines' component…
- How can you ensure that an OpenShift AI workbench always starts on a dedicated machine learning node pool ins…
- Which runtime dependency requires a cluster admin to install the Node Feature Discovery (NFD) Operator?
- Which component is used to view real-time graphical analytics and training performance graphs such as scalar…
- What must be defined in the Data Connection to connect an OpenShift AI workbench to a private Ceph Object Gat…
- What happens if you delete a Data Science Project directly from the OpenShift AI dashboard interface?
Answers and explanations
Tap a question to expand the answer and the exam reasoning. Try to commit to your own pick first.
Q1. Which components are deployed by default when the DataScienceCluster resource sets the 'pipelines' component state to 'Managed'?
Answer: C. Tekton pipelines architecture and APIServer components
OpenShift AI pipelines rely on the Tekton architecture and APIServer components. Tools like Jenkins or Argo Workflows are incorrect distractors not deployed by the OpenShift AI DataScienceCluster resource.
Q2. How can you ensure that an OpenShift AI workbench always starts on a dedicated machine learning node pool inside a large enterprise cluster?
Answer: C. Configure node selectors or tolerations within the workbench settings or cluster project parameters
Configuring node selectors or tolerations targets workbenches to specific machine learning nodes. Manually assigning pod IP addresses or changing control plane passwords will not automate workload placement.
Q3. Which runtime dependency requires a cluster admin to install the Node Feature Discovery (NFD) Operator?
Answer: B. NVIDIA GPU Acceleration support
The Node Feature Discovery Operator labels hardware nodes to enable NVIDIA GPU acceleration support. Standard storage provisioning and basic network separation do not require hardware discovery.
Q4. Which component is used to view real-time graphical analytics and training performance graphs such as scalar metrics directly from workbenches?
Answer: B. TensorBoard component
The TensorBoard component visualizes real-time machine learning metrics directly from your workbench. Grafana dashboards and OpenShift Logging track infrastructure health rather than model training performance.
Q5. What must be defined in the Data Connection to connect an OpenShift AI workbench to a private Ceph Object Gateway?
Answer: A. Endpoint URL, Access Key, and Secret Key
Private object storage connections require an endpoint URL, access key, and secret key. Container registry credentials authenticate image pulls, while node selectors manage workload placement, not storage access.
Q6. What happens if you delete a Data Science Project directly from the OpenShift AI dashboard interface?
Answer: D. The underlying OpenShift namespace and all its resources are completely removed
Deleting a Data Science Project from the dashboard removes the corresponding OpenShift namespace and all associated resources. Beware of options suggesting the workbench is deleted while the namespace remains; deletion is comprehensive at the project level.
Q7. Which runtime configuration must match the model framework when configuring ModelMesh serving?
Answer: D. Serving Runtime definition
The Serving Runtime definition must match your model framework, such as ONNX or PyTorch, to deploy properly. Network plugins and container runtime sockets are infrastructure-level configurations unrelated to model serving compatibility.
Q8. Which environment variable or setting determines the idle timeout after which a data science workbench is automatically stopped?
Answer: D. Notebook Idle Culling
Notebook Idle Culling determines the idle timeout after which a data science workbench is automatically stopped. Watch out for distractors like Auto-Stop or Pod Lifecycle MaxAge, which are not valid dashboard configuration terms.
Q9. Which custom resource configuration determines the maximum container size options shown in the dashboard?
Answer: C. OdhDashboardConfig definition settings profiles
The OdhDashboardConfig custom resource defines settings like container size profiles available in the dashboard. Standard cluster resource quotas impose actual limits, but they do not populate the menu selections in the interface.
Q10. Which log analytics tool is natively integrated to view historical logs of served models?
Answer: B. OpenShift Logging (Vector/Loki or Fluentd/ElasticSearch)
OpenShift Logging, using tools like Loki or Elasticsearch, is natively integrated to view historical logs of served models. Avoid choosing tracing systems like Jaeger, which monitor request paths and performance metrics rather than application logs.
Q11. When creating an Elyra pipeline, what file format defines the pipeline structure visually before it is submitted for execution?
Answer: A. .pipeline
Elyra uses a proprietary pipeline format with a dot pipeline extension to visually define the directed acyclic graph before exporting it. While you can ultimately export the file as YAML for Tekton runtime execution, the visual definition is strictly dot pipeline.
Q12. Which storage class type is strongly recommended for OpenShift AI notebook persistent volume claims (PVCs) to allow individual user data persistence?
Answer: A. ReadWriteOnce (RWO)
ReadWriteOnce is the correct access mode for a standard user notebook because it ensures dedicated data persistence for a single workbench instance. ReadWriteMany is typically reserved for distributed storage like shared pipeline artifacts.
Q13. What setting inside a Workbench CR controls the automatic termination of the workbench session due to prolonged inactivity?
Answer: A. Notebook runtime idle timeout configuration annotations or associated operator configurations
Workbench session culling is managed by the notebook controller via specific idle timeout annotations applied directly to the custom resource. Standard cluster pod eviction parameters will not successfully trigger culling for idle Jupyter interfaces.
Q14. How can an administrator monitor the specific performance and health of OpenShift AI operators?
Answer: A. Through the OpenShift Web Console Alerting and Metrics page
The OpenShift Web Console Alerting page monitors operator health and cluster metrics. Opening a bash session is a troubleshooting step, but it is not the primary monitoring mechanism.
Q15. What setting inside the DataScienceCluster resource disables a component completely so the operator stops managing or deploying it?
Answer: D. Removed
Setting a component to Removed in the DataScienceCluster resource fully disables it, causing the operator to stop managing and deploying it. Remember that Disabled merely pauses management, whereas Removed deletes the deployment.
Q16. When creating an object storage data connection, which parameter is required alongside the access key?
Answer: A. Secret Key
An object storage data connection requires both an access key and a secret key. Do not confuse this with AWS IAM roles or cluster identifiers, which are not used for standard S3-compatible connections.
Q17. Which open-source project provides the core distributed computing capabilities for execution steps inside an OpenShift AI Ray cluster?
Answer: D. Ray Core framework orchestration engine
OpenShift AI Ray clusters rely on the Ray Core framework for distributed computing and orchestration. Be careful not to confuse Ray with Apache Spark or Kubernetes batch plugins, which are different technologies.
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.