Nvidia-smi Tool Alert - Troubleshooting Guide
Purpose
This document provides a standardized approach for handling and troubleshooting the Nvidia-smi Tool alert, including L1 and L2 actions, escalation guidelines, and resolution criteria.
Alert Name
Nvidia-smi Tool
Alert Description
This alert is triggered when the nvidia-smi command fails to execute successfully on a node.
The alert indicates a potential issue with:
NVIDIA driver stack
GPU communication
System-level
GPU services or state
Failure of this command may result in:
-
Inability to monitor GPU health
- GPU workloads being impacted
- Loss of visibility into GPU utilization and errors
Severity
P3 – Medium Impact
- No immediate outage, but GPU observability or performance may be affected
-
Requires timely investigation to prevent escalation
Possible Causes
- NVIDIA drivers not installed or corrupted
- GPU not detected by OS
- NVIDIA persistence daemon not running
- Kernel module issues
- Hardware or PCIe communication failure
- Recent system changes or updates
L1 Engineer Actions
Step 1 – Log the Alert
- Log in to the affected node
- Validate the alert
- Wait for 15-20 minutes if alert gets auto resolved or false alarm.
Step 2 – Create Ticket
- Verify if the issue persists
- Capture error output (if any)
- Create an internal ticket with:
- Error output
- Observations
- Node details
Step 3 – Escalation (If required)
- Inform L2 team immediately
- Share all collected details and logs
Step 4 – Documentation
- Update Alert Summary with:
- Observations
- Error messages
- After L2 confirmation, update:
- Issue Master Sheet / Issue Tracking Sheet
L2 Engineer Actions
Step 1 – Check NVIDIA Persistence Service
systemctl status nvidia-persistenced
- Ensure the service is active (running)
- Restart if required
systemctl restart nvidia-persistenced
Step 2 – Verify NVIDIA Driver Status
nvidia-smi
- If still failing:
lsmod | grep nvidia
- Confirm NVIDIA kernel modules are loaded
- Check for driver mismatch or failure
Step 3 – Check System Logs
dmesg | grep -i nvidia
journalctl -xe | grep -i nvidia
- Identify driver errors or hardware-related issues.
Step 4 – Validate GPU Detection
lspci | grep -i nvidia
- Ensure GPU is visible at hardware level.
Step 5 – OEM Escalation (If required)
- If issue persists after troubleshooting:
- Create a ticket with OEM
- Share logs, output, and observations.
Resolution
The issue is considered resolved when:
-
nvidia-smi executes successfully
-
GPU details are displayed correctly
-
No related errors are observed in system logs
Escalation
- Internal Escalation: L1 → L2 with complete logs and observations
- External Escalation: L2 → OEM (if hardware/driver issue persists
Related Articles
GPU Miscount Alert - Troubleshooting Guide
Purpose Provide troubleshooting steps when a GPU Miscount alert is triggered, indicating that the system detects an incorrect number of GPUs on a node. Alert Name GPU Miscount What the Alert Is This alert indicates that the number of GPUs detected on ...
GPU Memory Mismatch – Troubleshooting Guide
Alert Information Alert Name: GPU Memory Mismatch Severity: P3 – Medium Impact Alert Description: This alert is generated when the detected GPU memory size on a node does not match the expected configuration. Alert Summary The monitoring system has ...
Execution of NVIDIA Field Diagnostic (FD) Tool and Collection of Diagnostic Logs, and troubleshooting of GPU(s) not detected
1. Objective To provide a standardized procedure for executing the NVIDIA Field Diagnostic (FD) tool, verifying GPU status, collecting the required diagnostic logs, and documenting the findings for further analysis. 2. Scope This procedure applies to ...
GPU Memory Temperature Critical - Troubleshooting Guide
Purpose This document provides a standardized approach for handling and troubleshooting the GPU Memory Temperature Critical alert, including L1 and L2 actions, escalation procedures, and resolution criteria. Alert Name GPU Memory Temperature Critical ...
Network Down Extern (P1 Critical Alert) - Troubleshooting Guide
Purpose This document provides a standardized approach for handling and troubleshooting the Network Down - Extern alert, including L1 and L2 actions, escalation procedures, and resolution criteria for critical incidents. Alert Name Network Down - ...