NCP-AIO 試験問題を無料オンラインアクセス

A system administrator needs to scale a Kubernetes Job to 4 replicas.
What command should be used?

• A.kubectl scale job --replicas=4
• B.kubectl scale job -r 4
• C.kubectl stretch job --replicas=4
• D.kubectl autoscale deployment job --min=1 --max=10

You are an administrator managing a large-scale Kubernetes-based GPU cluster using Run:AI.
To automate repetitive administrative tasks and efficiently manage resources across multiple nodes, which of the following is essential when using the Run:AI Administrator CLI for environments where automation or scripting is required?

• A.Install the CLI on Windows machines to take advantage of its scripting capabilities.
• B.Ensure that the Kubernetes configuration file is set up with cluster administrative rights before using the CLI.
• C.Use the CLI to manually allocate specific GPUs to individual jobs for better resource management.
• D.Use the runai-adm command to directly update Kubernetes nodes without requiring kubectl.

You need to do maintenance on a node. What should you do first?

• A.Set the node state to down in Slurm before completing maintenance.
• B.Drain the compute node using scontrol update.
• C.Set the node state to down in Slurm before completing maintenance.
• D.Disable job scheduling on all compute nodes in Slurm before completing maintenance.

Your organization is deploying an AI workload that requires high-throughput access to shared storage across multiple servers. The workload involves both training and inference tasks that need fast read and write speeds.
Which storage architecture would best support this AI workload?

• A.Prioritize write performance over read performance since training tasks dominate AI workflows.
• B.Use SSD-based shared storage systems to save costs while scaling up storage capacity.
• C.A high-performance shared storage system that supports both high read and write IO performance.
• D.Use local storage on each server to minimize network traffic between nodes.

In a high availability (HA) cluster, you need to ensure that split-brain scenarios are avoided.
What is a common technique used to prevent split-brain in an HA cluster?

• A.Replicating data across all nodes in real time.
• B.Configuring manual failover procedures for each node.
• C.Using multiple load balancers to distribute traffic evenly across nodes.
• D.Implementing a heartbeat network between cluster nodes to monitor their health.