Red Hat Certified Specialist in OpenShift AI (EX267) Practice Exam Questions and – Part 61/65

Answer: C. ImageStreams

ImageStreams are the underlying OpenShift resources used to register and expose custom notebook images in the dashboard. On the exam, remember that the OpenShift AI dashboard simply queries these cluster resources to populate its workbench image dropdown list.

Answer: D. Open Data Hub

Open Data Hub serves as the upstream community project for Red Hat OpenShift AI. Kubeflow is related but distinct, so remember that OpenShift AI is built directly on top of the Open Data Hub operator and its components.

Answer: A. Red Hat OpenShift Data Foundation (ODF) / Ceph object storage

Red Hat OpenShift Data Foundation, utilizing Ceph, provides the default S3-compatible object storage for on-premise pipeline artifacts. Rely on ODF for internal storage configurations during the exam rather than cloud-specific or block storage options.

Answer: D. Model Registry component

The Model Registry component tracks model versions, states, and registry locations within OpenShift AI. Tekton handles pipeline tasks but does not serve as the centralized repository for machine learning model metadata and version histories.

Answer: D. Environment Variables from Secrets

Environment variables populated from Kubernetes Secrets securely pass credentials to workbenches without exposing them in plain text. Hardcoded ConfigMaps fail as a distractor because they store non-sensitive data publicly within the namespace.

Answer: B. OpenShift OAuth Server

The OpenShift OAuth Server handles authentication for the dashboard by relying directly on the cluster's built-in identity providers. Red Hat SSO is a distractor because it is not the default authentication mechanism for OpenShift AI.

Answer: B. OpenShift OAuth proxy container

The OpenShift OAuth proxy container intercepts dashboard requests to validate user tokens against the cluster identity provider. Red Hat Insights and Active Directory are distractors, while local file hashes do not handle network authentication.

Answer: B. oc

The oc command line tool is the primary interface for managing OpenShift clusters. While kubectl works on Kubernetes resources, oc provides the specific capabilities needed for OpenShift administration.

Answer: D. ReadWriteMany (RWX) PV supported by enterprise file storage

ReadWriteMany (RWX) is the correct access mode because it allows multiple pods to mount and write to the same persistent volume simultaneously. The other options restrict access to a single node or use temporary storage unsuitable for scalable workbenches.

Answer: D. AWS_SECRET_ACCESS_KEY

Data Connections in OpenShift AI store S3 credentials in a Kubernetes Secret using the standard AWS environment variable naming convention. The AWS_SECRET_ACCESS_KEY variable is automatically exposed to the workbench or application, ensuring seamless SDK authentication.

Answer: D. Route

An OpenShift Route exposes an internal Service externally via HTTPS, allowing outside web traffic to reach the deployed model. Standard Kubernetes Ingress is not the primary routing mechanism in OpenShift, making Route the required resource.

Answer: B. Data Science Projects > Models and model servers

The Data Science Projects section contains the Models and model servers tab where users manage deployments. From here, users can view runtime performance metrics, access logs, and monitor the active status of their served machine learning models.

Answer: C. Configuring appropriate tolerations and node selectors matching the hardware taint labels

Applying tolerations and node selectors ensures the scheduler places the workbench pod on nodes with specialized hardware. This configuration overrides taints designed to repel standard workloads, guaranteeing access to the required accelerators.

Answer: A. They are read from the global cluster proxy configuration via DSCInitialization

The DSCInitialization custom resource automatically reads the OpenShift cluster-wide proxy configuration to inject proxy settings into OpenShift AI components. Avoid manual workbench configuration, as the operator handles this globally during initialization to ensure proper network routing.

Answer: D. Scaling the replicas of the model serving pods across multiple availability zones and nodes

High availability is achieved by scaling the number of model serving replicas across different nodes and availability zones. Relying on a single large server provides no redundancy, meaning any pod failure disrupts the model inference endpoint.

Answer: B. Elyra component runtime wrapper

The Elyra extension provides a visual interface to convert standard Python scripts into containerized pipeline tasks. KServe is used for model serving, not authoring pipelines, making it an incorrect choice for workflow generation.

Answer: B. Red Hat Quay, Docker Hub, or any secure OCI-compliant registry

OpenShift AI pulls custom notebook images from any secure OCI-compliant registry, including Quay and Docker Hub. Git repositories only store source code, not the built container images required by the cluster to launch workbenches.

Answer: C. It is sent securely to Red Hat unless explicitly disabled by an administrator

OpenShift AI sends anonymized usage telemetry data to Red Hat by default unless an administrator explicitly disables it through the operator configuration. It is not locked behind a premium license, nor is it permanently restricted to local storage.