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 is the standard name of the custom resource definition used to control global settings for the OpenShift AI initial. 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 is the standard name of the custom resource definition used to control global settings for the OpenShift…
- Which component must be configured within an OpenShift AI project before you can import or run any pipeline w…
- Which role-based access control (RBAC) permission must a user have within an OpenShift namespace to view and…
- When configuring KServe, which component manages ingress routing and provides network isolation between diffe…
- What is the minimum recommended number of worker nodes required to install the OpenShift AI Operator in a sta…
- What type of API protocol is universally supported by OpenShift AI serving runtimes for sending inference req…
Answers and explanations
Tap a question to expand the answer and the exam reasoning. Try to commit to your own pick first.
Q1. What is the standard name of the custom resource definition used to control global settings for the OpenShift AI initialization phase?
Answer: B. DSCInitialization
The DSCInitialization custom resource definition controls global settings during the OpenShift AI initialization phase. It handles foundational platform prerequisites before the main components are deployed.
Q2. Which component must be configured within an OpenShift AI project before you can import or run any pipeline workflows?
Answer: A. An object storage Data Connection and a Pipeline Server
You must configure an object storage Data Connection and a Pipeline Server before running workflows in OpenShift AI. The Pipeline Server requires external storage to persist the pipeline artifacts and outputs.
Q3. Which role-based access control (RBAC) permission must a user have within an OpenShift namespace to view and execute pipelines but not modify project settings?
Answer: C. view
The basic view role provides read-only access to cluster resources, allowing users to inspect pipelines without modifying project settings. The edit role is a distractor because it grants write permissions to alter resources.
Q4. When configuring KServe, which component manages ingress routing and provides network isolation between different models?
Answer: B. OpenShift Service Mesh / Istio Ingress Gateway
KServe relies on the Istio Ingress Gateway provided by OpenShift Service Mesh for intelligent routing and network isolation between deployed models. Standard OpenShift Routes do not provide the advanced traffic splitting required.
Q5. What is the minimum recommended number of worker nodes required to install the OpenShift AI Operator in a standard self-managed cluster?
Answer: D. 2 worker nodes
A minimum of two worker nodes is recommended to ensure high availability and provide adequate compute resources for the OpenShift AI components. A single worker node lacks the redundancy required for a stable deployment.
Q6. What type of API protocol is universally supported by OpenShift AI serving runtimes for sending inference requests to deployed models?
Answer: A. gRPC and HTTP/REST
OpenShift AI serving runtimes accept standard inference requests via gRPC and HTTP/REST protocols. The other options are legacy communication protocols and are not used for model inference, making them straightforward distractors.
Q7. What is the underlying engine used by OpenShift AI for Single Model Serving?
Answer: A. KServe
KServe is the underlying custom resource engine utilized by OpenShift AI to deploy models via single-model serving. ModelMesh is used for multi-model serving, while TorchServe and Triton function as specific runtime containers rather than the management engine.
Q8. What format is used to write pipelines manually using the OpenShift AI Pipelines SDK?
Answer: D. Python code leveraging the Tekton/KFP SDK packages
Pipelines are written manually using Python code that leverages the Kubeflow Pipelines or Tekton SDK packages. OpenShift AI relies on these Python SDKs rather than XML, shell scripts, or SQL to define pipeline directed acyclic graphs.
Q9. Which tool provides dashboards for visual analysis of training runs and performance inside the workbench?
Answer: D. TensorBoard
TensorBoard is the standard visualizer bundled with Jupyter workbenches for examining metrics like loss and accuracy. Grafana mainly targets cluster operators for infrastructure monitoring, not model training metrics inside a notebook.
Q10. Which resource allows users to create reproducible workflows containing multiple execution steps?
Answer: C. Data Science Pipeline
Data Science Pipelines allow users to build automated, reproducible machine learning workflows with multiple steps. While Elyra helps visually assemble these pipelines, the pipeline resource itself is what actually executes the workload.
Q11. What error occurs if a workbench requests a GPU but the cluster has no available GPU nodes or allocatable resources?
Answer: B. Pod status remains in 'Pending' state
If a GPU is requested but unavailable, the Kubernetes scheduler cannot bind the pod to a node, leaving it in a Pending state. OpenShift AI never silently downgrades workbench deployments to CPU automatically.
Q12. Which internal object handles the synchronization and loading of model files from S3 storage into a Multi-Model server runtime?
Answer: B. Storage-initializer container sidecar
The storage-initializer sidecar container pulls the model files from the S3 connection into the runtime pod's local directory. OpenShift Routes and standard Kubelet daemons do not handle pulling model files from object storage.
Q13. How can an administrator change the global default storage size allocated to newly created workbench storage volumes?
Answer: D. By modifying the settings in the OpenShift AI dashboard configuration or cluster storage class
Storage size is typically configured through cluster storage classes or dashboard settings. However, the provided options mix unrelated infrastructure configurations with valid settings, making this question ambiguous and poorly structured.
Q14. Which CLI tool is required to manage pipeline definitions directly via the terminal interface?
Answer: A. tkn or kfp CLI utility
The OpenShift Pipelines CLI or the Kubeflow Pipelines SDK manages pipeline definitions via terminal. Providing both as a single option introduces slight ambiguity. Remember that rh-ai and oc-ai are distractors.
Q15. Which open-source server is commonly used by OpenShift AI to serve ONNX and predictive models under ModelMesh?
Answer: B. OVMS (OpenVINO Model Server) or Triton Inference Server
The OpenVINO Model Server acts as the standard runtime within ModelMesh for handling ONNX and predictive models. Nginx and RabbitMQ handle web traffic and messaging rather than executing machine learning inference workloads.
Q16. How can an administrator inject a custom environment variable globally into all newly created data science workbenches?
Answer: A. By modifying the DataScienceCluster custom resource
The DataScienceCluster custom resource acts as the central configuration hub for OpenShift AI components. The other options are overly complex or invalid because the dashboard and operator configuration directly manage global deployments without needing custom admission webhooks.
Q17. What component handles the aggregation and rendering of runtime metrics for models served via ModelMesh?
Answer: A. Prometheus and Grafana integration via the OpenShift Cluster Monitoring stack
OpenShift AI relies on the native OpenShift Cluster Monitoring stack, which uses Prometheus to collect metrics and Grafana to render dashboards. Custom logging pipelines or direct etcd queries are not supported for model metrics.
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.