cert
cert-1
cert-2

Pass NVIDIA NCP-AII Exam in First Attempt Guaranteed!

Get 100% Latest Exam Questions, Accurate & Verified Answers to Pass the Actual Exam!
30 Days Free Updates, Instant Download!

cert-5
cert-6
NCP-AII Exam - Verified By Experts
NCP-AII Premium File

NCP-AII Premium File

$89.99
$98.99
  • Premium File 170 Questions & Answers. Last Update: Nov 06, 2026

Whats Included:

  • Latest Questions
  • 100% Accurate Answers
  • Fast Exam Updates
 
$98.99
$89.99
accept 14 downloads in the last 7 days
block-screenshots
NVIDIA NCP-AII Practice Test 44

Question 430

During an NVIDIA AI factory acceptance test, engineers discover: A cluster's scheduler advertises full physical GPUs while the nodes expose only MIG partition devices. Which TWO complementary elements are needed: the correct technical action and the strongest independent diagnostic evidence? Select TWO answers.

  1. Choose supported MIG GPU-instance profiles and create matching compute instances with validated resource assignments.
  2. Coordinate MIG reconfiguration with workload draining, persistent desired-state recreation and scheduler resource discovery.
  3. Inspect running process ownership, postboot instance layout, device plugin or GRES inventory and schedule placement receipts.
  4. Inspect DPU hardware generation, BFB package provenance, BMC job records, Arm boot logs and firmware versions.
  5. Treat MIG slice identifiers and scheduling availability as immutable across resets without checking instance recreation.

Explanation

Correct Answers

B:

This is an appropriate correction for MIG lifecycle and workload scheduling persistence: Coordinate MIG reconfiguration with workload draining, persistent desired-state recreation and scheduler resource discovery. Observed scenario: A cluster's scheduler advertises full physical GPUs while the nodes expose only MIG partition devices. The governing mechanism is: MIG instance topology and scheduler-visible resources must agree; some MIG instance state requires explicit recreation after reboot.

C:

This is the direct operational evidence to investigate: Inspect running process ownership, postboot instance layout, device plugin or GRES inventory and schedule placement receipts. Observed scenario: A cluster's scheduler advertises full physical GPUs while the nodes expose only MIG partition devices.

Incorrect Answers

A:

This alternative is relevant to MIG GPU and compute instance partition profiles but does not directly resolve the current MIG lifecycle and workload scheduling persistence issue: Choose supported MIG GPU-instance profiles and create matching compute instances with validated resource assignments. Observed scenario: A cluster's scheduler advertises full physical GPUs while the nodes expose only MIG partition devices. The governing mechanism is: MIG instance topology and scheduler-visible resources must agree; some MIG instance state requires explicit recreation after reboot.

D:

This alternative is relevant to BlueField boot image BFB provisioning and BMC but does not directly resolve the current MIG lifecycle and workload scheduling persistence issue: Inspect DPU hardware generation, BFB package provenance, BMC job records, Arm boot logs and firmware versions. Observed scenario: A cluster's scheduler advertises full physical GPUs while the nodes expose only MIG partition devices.

E:

This tempting but unsafe shortcut is not a sufficient remediation for MIG lifecycle and workload scheduling persistence: Treat MIG slice identifiers and scheduling availability as immutable across resets without checking instance recreation. A cluster's scheduler advertises full physical GPUs while the nodes expose only MIG partition devices. MIG instance topology and scheduler-visible resources must agree; some MIG instance state requires explicit recreation after reboot.

Question 431

A production GPU cluster review identifies this condition: A cluster's scheduler advertises full physical GPUs while the nodes expose only MIG partition devices. Which diagnostic evidence should be obtained FIRST to substantiate the likely fault or configuration gap? Select ONE best answer.

  1. Inspect mode configuration, embedded OS status, host-visible functions, network policy and mode-transition logs.
  2. Treat MIG slice identifiers and scheduling availability as immutable across resets without checking instance recreation.
  3. Inspect running process ownership, postboot instance layout, device plugin or GRES inventory and schedule placement receipts.
  4. Inspect GPU model and MIG support, available profile geometry, GPU instance and compute instance listings and CUDA visibility.

Explanation

Correct Answer

C:

This is the direct operational evidence to investigate: Inspect running process ownership, postboot instance layout, device plugin or GRES inventory and schedule placement receipts. Observed scenario: A cluster's scheduler advertises full physical GPUs while the nodes expose only MIG partition devices. The governing mechanism is: MIG instance topology and scheduler-visible resources must agree; some MIG instance state requires explicit recreation after reboot.

Incorrect Answers

A:

This alternative is relevant to BlueField DPU embedded versus NIC ownership but does not directly resolve the current MIG lifecycle and workload scheduling persistence issue: Inspect mode configuration, embedded OS status, host-visible functions, network policy and mode-transition logs. Observed scenario: A cluster's scheduler advertises full physical GPUs while the nodes expose only MIG partition devices.

B:

This tempting but unsafe shortcut is not a sufficient remediation for MIG lifecycle and workload scheduling persistence: Treat MIG slice identifiers and scheduling availability as immutable across resets without checking instance recreation. Observed scenario: A cluster's scheduler advertises full physical GPUs while the nodes expose only MIG partition devices. The governing mechanism is: MIG instance topology and scheduler-visible resources must agree; some MIG instance state requires explicit recreation after reboot. The proportionate action is: Coordinate MIG reconfiguration with workload draining, persistent desired-state recreation and scheduler resource discovery.

D:

This alternative is relevant to MIG GPU and compute instance partition profiles but does not directly resolve the current MIG lifecycle and workload scheduling persistence issue: Inspect GPU model and MIG support, available profile geometry, GPU instance and compute instance listings and CUDA visibility. Observed scenario: A cluster's scheduler advertises full physical GPUs while the nodes expose only MIG partition devices.

Question 432

In a professional AI infrastructure deployment, a technician observes: A cluster's scheduler advertises full physical GPUs while the nodes expose only MIG partition devices. After corrective work, which observable result BEST confirms that the requirement is met? Select ONE best answer.

  1. After maintenance the expected GPU slices are recreated and queued workloads receive only valid assigned devices.
  2. The DPU boots its expected software, responds on approved management paths and exposes intended host functions.
  3. Host and DPU traffic follows the intended control model after restart, with reachable protected management access.
  4. Treat MIG slice identifiers and scheduling availability as immutable across resets without checking instance recreation.

Explanation

Correct Answer

A:

This is an observable verified result, not merely a claimed configuration change: After maintenance the expected GPU slices are recreated and queued workloads receive only valid assigned devices. Observed scenario: A cluster's scheduler advertises full physical GPUs while the nodes expose only MIG partition devices.

Incorrect Answers

B:

This alternative is relevant to BlueField boot image BFB provisioning and BMC but does not directly resolve the current MIG lifecycle and workload scheduling persistence issue: The DPU boots its expected software, responds on approved management paths and exposes intended host functions. Observed scenario: A cluster's scheduler advertises full physical GPUs while the nodes expose only MIG partition devices.

C:

This alternative is relevant to BlueField DPU embedded versus NIC ownership but does not directly resolve the current MIG lifecycle and workload scheduling persistence issue: Host and DPU traffic follows the intended control model after restart, with reachable protected management access. Observed scenario: A cluster's scheduler advertises full physical GPUs while the nodes expose only MIG partition devices.

D:

This tempting but unsafe shortcut is not a sufficient remediation for MIG lifecycle and workload scheduling persistence: Treat MIG slice identifiers and scheduling availability as immutable across resets without checking instance recreation. Observed scenario: A cluster's scheduler advertises full physical GPUs while the nodes expose only MIG partition devices. The governing mechanism is: MIG instance topology and scheduler-visible resources must agree; some MIG instance state requires explicit recreation after reboot. The proportionate action is: Coordinate MIG reconfiguration with workload draining, persistent desired-state recreation and scheduler resource discovery. A direct check is: Inspect running process ownership, postboot instance layout, device plugin or GRES inventory and schedule placement receipts. The acceptance condition is: After maintenance the expected GPU slices are recreated and queued workloads receive only valid assigned devices.

Question 433

The cluster operations team needs to evaluate this situation: A new cluster installs compute-node images without a functioning BCM management node or provisioning services. Which operational response is MOST appropriate to address the root issue? Select ONE best answer.

  1. Install a supported NGC CLI, use scoped secret handling and retrieve versioned registry artifacts reproducibly.
  2. Install the supported BCM head-node services and preserve configuration data before provisioning and accepting workers.
  3. Assume a collection of individually installed Linux servers is a fully managed cluster without working BCM services.
  4. Define supported Slurm controller, compute nodes, partitions and authorized scheduling policies together.

Explanation

Correct Answer

B:

This is an appropriate correction for Base Command Manager head-node installation: Install the supported BCM head-node services and preserve configuration data before provisioning and accepting workers. Observed scenario: A new cluster installs compute-node images without a functioning BCM management node or provisioning services.

Incorrect Answers

A:

That option concerns NGC CLI installation login and artifact retrieval. Cluster managers coordinate images, nodes, services and configuration; worker OS boot cannot replace control-plane readiness.

C:

This tempting but unsafe shortcut is not a sufficient remediation for Base Command Manager head-node installation: Assume a collection of individually installed Linux servers is a fully managed cluster without working BCM services.

D:

This alternative is relevant to Slurm controller partition setup but does not directly resolve the current Base Command Manager head-node installation issue: Define supported Slurm controller, compute nodes, partitions and authorized scheduling policies together. Observed scenario: A new cluster installs compute-node images without a functioning BCM management node or provisioning services.

Question 434

At a multi-node AI infrastructure handoff, the team records: A new cluster installs compute-node images without a functioning BCM management node or provisioning services. Which underlying infrastructure principle BEST explains the issue? Select ONE best answer.

  1. Assume a collection of individually installed Linux servers is a fully managed cluster without working BCM services.
  2. Cluster managers coordinate images, nodes, services and configuration; worker OS boot cannot replace control-plane readiness.
  3. IP connectivity does not prove correct RDMA capability, link aggregation or loss and congestion treatment.
  4. Host administration requires least privilege and credential lifecycle control independently of GPU job permissions.

Explanation

Correct Answer

B:

This is the governing infrastructure principle: Cluster managers coordinate images, nodes, services and configuration; worker OS boot cannot replace control-plane readiness. Observed scenario: A new cluster installs compute-node images without a functioning BCM management node or provisioning services.

Incorrect Answers

A:

This tempting but unsafe shortcut is not a sufficient remediation for Base Command Manager head-node installation: Assume a collection of individually installed Linux servers is a fully managed cluster without working BCM services. Cluster managers coordinate images, nodes, services and configuration; worker OS boot cannot replace control-plane readiness.

C:

That option concerns Compute fabric interface bonding and RDMA prerequisites. A direct check is: Inspect BCM service health, database backups, node discovery, management interfaces and provisioning records.

D:

This alternative is relevant to Host SSH certificates keys and OS access segregation but does not directly resolve the current Base Command Manager head-node installation issue: Host administration requires least privilege and credential lifecycle control independently of GPU job permissions. Observed scenario: A new cluster installs compute-node images without a functioning BCM management node or provisioning services.

Question 435

During an NVIDIA AI factory acceptance test, engineers discover: A new cluster installs compute-node images without a functioning BCM management node or provisioning services. Which diagnostic evidence should be obtained FIRST to substantiate the likely fault or configuration gap? Select ONE best answer.

  1. Inspect BCM service health, database backups, node discovery, management interfaces and provisioning records.
  2. Review authorized key and account policy, SSH server settings, firewall reachability and authentication logs.
  3. Assume a collection of individually installed Linux servers is a fully managed cluster without working BCM services.
  4. Inspect DHCP leases, PXE requests, image revisions, BCM node mappings and host boot traces.

Explanation

Correct Answer

A:

This is the direct operational evidence to investigate: Inspect BCM service health, database backups, node discovery, management interfaces and provisioning records. Observed scenario: A new cluster installs compute-node images without a functioning BCM management node or provisioning services.

Incorrect Answers

B:

This alternative is relevant to Host SSH certificates keys and OS access segregation but does not directly resolve the current Base Command Manager head-node installation issue: Review authorized key and account policy, SSH server settings, firewall reachability and authentication logs. Observed scenario: A new cluster installs compute-node images without a functioning BCM management node or provisioning services.

C:

This tempting but unsafe shortcut is not a sufficient remediation for Base Command Manager head-node installation: Assume a collection of individually installed Linux servers is a fully managed cluster without working BCM services. A new cluster installs compute-node images without a functioning BCM management node or provisioning services.

D:

This alternative is relevant to BCM image provisioning and PXE worker boot but does not directly resolve the current Base Command Manager head-node installation issue: Inspect DHCP leases, PXE requests, image revisions, BCM node mappings and host boot traces. Observed scenario: A new cluster installs compute-node images without a functioning BCM management node or provisioning services.

Question 436

A production GPU cluster review identifies this condition: A new cluster installs compute-node images without a functioning BCM management node or provisioning services. After corrective work, which observable result BEST confirms that the requirement is met? Select ONE best answer.

  1. Correct GPU workers receive the intended profile without unrelated node categories being modified.
  2. Assume a collection of individually installed Linux servers is a fully managed cluster without working BCM services.
  3. The management node reports correct cluster state and consistently deploys an approved image to intended nodes.
  4. The requested test job enters the correct partition and runs on an eligible healthy GPU node.

Explanation

Correct Answer

C:

This is an observable verified result, not merely a claimed configuration change: The management node reports correct cluster state and consistently deploys an approved image to intended nodes. Observed scenario: A new cluster installs compute-node images without a functioning BCM management node or provisioning services.

Incorrect Answers

A:

This alternative is relevant to BCM node categories and role inheritance but does not directly resolve the current Base Command Manager head-node installation issue: Correct GPU workers receive the intended profile without unrelated node categories being modified. Observed scenario: A new cluster installs compute-node images without a functioning BCM management node or provisioning services.

B:

This tempting but unsafe shortcut is not a sufficient remediation for Base Command Manager head-node installation: Assume a collection of individually installed Linux servers is a fully managed cluster without working BCM services. A new cluster installs compute-node images without a functioning BCM management node or provisioning services. Cluster managers coordinate images, nodes, services and configuration; worker OS boot cannot replace control-plane readiness.

D:

This alternative is relevant to Slurm controller partition setup but does not directly resolve the current Base Command Manager head-node installation issue: The requested test job enters the correct partition and runs on an eligible healthy GPU node. Observed scenario: A new cluster installs compute-node images without a functioning BCM management node or provisioning services.

Question 437

In a professional AI infrastructure deployment, a technician observes: An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware. Which operational response is MOST appropriate to address the root issue? Select ONE best answer.

  1. Assume a collection of individually installed Linux servers is a fully managed cluster without working BCM services.
  2. Define supported Slurm controller, compute nodes, partitions and authorized scheduling policies together.
  3. Validate DHCP/PXE services, provisioned image and node identity mapping, then test clean node reboots.
  4. Install the supported BCM head-node services and preserve configuration data before provisioning and accepting workers.

Explanation

Correct Answer

D:

This is an appropriate correction for Base Command Manager head-node installation: Install the supported BCM head-node services and preserve configuration data before provisioning and accepting workers. Observed scenario: An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware.

Incorrect Answers

A:

This tempting but unsafe shortcut is not a sufficient remediation for Base Command Manager head-node installation: Assume a collection of individually installed Linux servers is a fully managed cluster without working BCM services. An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware.

B:

This alternative is relevant to Slurm controller partition setup but does not directly resolve the current Base Command Manager head-node installation issue: Define supported Slurm controller, compute nodes, partitions and authorized scheduling policies together. Observed scenario: An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware.

C:

This alternative is relevant to BCM image provisioning and PXE worker boot but does not directly resolve the current Base Command Manager head-node installation issue: Validate DHCP/PXE services, provisioned image and node identity mapping, then test clean node reboots. Observed scenario: An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware.

Question 438

The cluster operations team needs to evaluate this situation: An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware. Which underlying infrastructure principle BEST explains the issue? Select ONE best answer.

  1. Assume a collection of individually installed Linux servers is a fully managed cluster without working BCM services.
  2. A BCM category groups settings applied to member nodes; untracked manual changes may conflict with managed desired state.
  3. Cluster managers coordinate images, nodes, services and configuration; worker OS boot cannot replace control-plane readiness.
  4. A working Slurm daemon is not enough; partitions determine which nodes and resources can serve jobs.

Explanation

Correct Answer

C:

This is the governing infrastructure principle: Cluster managers coordinate images, nodes, services and configuration; worker OS boot cannot replace control-plane readiness. Observed scenario: An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware.

Incorrect Answers

A:

This tempting but unsafe shortcut is not a sufficient remediation for Base Command Manager head-node installation: Assume a collection of individually installed Linux servers is a fully managed cluster without working BCM services. An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware. Cluster managers coordinate images, nodes, services and configuration; worker OS boot cannot replace control-plane readiness.

B:

That option concerns BCM node categories and role inheritance, not the requirement in this case.

D:

This alternative is relevant to Slurm controller partition setup but does not directly resolve the current Base Command Manager head-node installation issue: A working Slurm daemon is not enough; partitions determine which nodes and resources can serve jobs. Observed scenario: An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware.

Question 439

At a multi-node AI infrastructure handoff, the team records: An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware. Which diagnostic evidence should be obtained FIRST to substantiate the likely fault or configuration gap? Select ONE best answer.

  1. Review authorized key and account policy, SSH server settings, firewall reachability and authentication logs.
  2. Inspect BCM service health, database backups, node discovery, management interfaces and provisioning records.
  3. Assume a collection of individually installed Linux servers is a fully managed cluster without working BCM services.
  4. Check HA roles, sync status, virtual management endpoint, database replication and isolation tests.

Explanation

Correct Answer

B:

This is the direct operational evidence to investigate: Inspect BCM service health, database backups, node discovery, management interfaces and provisioning records. Observed scenario: An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware.

Incorrect Answers

A:

This alternative is relevant to Host SSH certificates keys and OS access segregation but does not directly resolve the current Base Command Manager head-node installation issue: Review authorized key and account policy, SSH server settings, firewall reachability and authentication logs. Observed scenario: An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware.

C:

This tempting but unsafe shortcut is not a sufficient remediation for Base Command Manager head-node installation: Assume a collection of individually installed Linux servers is a fully managed cluster without working BCM services. Observed scenario: An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware. The governing mechanism is: Cluster managers coordinate images, nodes, services and configuration; worker OS boot cannot replace control-plane readiness. The proportionate action is: Install the supported BCM head-node services and preserve configuration data before provisioning and accepting workers.

D:

This alternative is relevant to BCM high availability and management failover but does not directly resolve the current Base Command Manager head-node installation issue: Check HA roles, sync status, virtual management endpoint, database replication and isolation tests. Observed scenario: An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware.

Question 440

During an NVIDIA AI factory acceptance test, engineers discover: An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware. Which TWO complementary elements are needed: the correct technical action and the strongest independent diagnostic evidence? Select TWO answers.

  1. Inspect slurmd discovery, gres configuration, CUDA_VISIBLE_DEVICES and simultaneous job placement.
  2. Assume a collection of individually installed Linux servers is a fully managed cluster without working BCM services.
  3. Install the supported BCM head-node services and preserve configuration data before provisioning and accepting workers.
  4. Inspect BCM service health, database backups, node discovery, management interfaces and provisioning records.
  5. Define explicit BCM node categories and managed overrides for shared hardware and service configuration.

Explanation

Correct Answers

C:

This is an appropriate correction for Base Command Manager head-node installation: Install the supported BCM head-node services and preserve configuration data before provisioning and accepting workers. Observed scenario: An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware. The governing mechanism is: Cluster managers coordinate images, nodes, services and configuration; worker OS boot cannot replace control-plane readiness.

D:

This is the direct operational evidence to investigate: Inspect BCM service health, database backups, node discovery, management interfaces and provisioning records. Observed scenario: An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware. The governing mechanism is: Cluster managers coordinate images, nodes, services and configuration; worker OS boot cannot replace control-plane readiness.

Incorrect Answers

A:

That option concerns Slurm GPU GRES enumeration and device mapping. Cluster managers coordinate images, nodes, services and configuration; worker OS boot cannot replace control-plane readiness.

B:

This tempting but unsafe shortcut is not a sufficient remediation for Base Command Manager head-node installation: Assume a collection of individually installed Linux servers is a fully managed cluster without working BCM services. Observed scenario: An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware. The governing mechanism is: Cluster managers coordinate images, nodes, services and configuration; worker OS boot cannot replace control-plane readiness. The proportionate action is: Install the supported BCM head-node services and preserve configuration data before provisioning and accepting workers. A direct check is: Inspect BCM service health, database backups, node discovery, management interfaces and provisioning records. The acceptance condition is: The management node reports correct cluster state and consistently deploys an approved image to intended nodes.

E:

This alternative is relevant to BCM node categories and role inheritance but does not directly resolve the current Base Command Manager head-node installation issue: Define explicit BCM node categories and managed overrides for shared hardware and service configuration. Observed scenario: An administrator starts workload scheduling before the cluster manager has discovered and accepted node hardware.

NCP-AI Infrastructure (NCP-AII) Practice Tests