GPU Memory Temperature Critical - Troubleshooting Guide

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

Alert Description

This alert is triggered when abnormally high temperature is detected in GPU memory modules.

GPU memory overheating can be more critical than GPU core temperature and may indicate:
  1. Improper airflow in the chassis
  2. High GPU workload
  3. Cooling system malfunction
  4. Hardware degradation or failure
If unresolved, this may lead to:
  1. GPU throttling
  2. Workload failures
  3. Hardware damage

Severity

P3 - Medium Impact

Possible Causes
  1. High GPU utilization due to workloads
  2. Improper chassis airflow
  3. Fan malfunction
  4. Data center cooling issues
  5. Dust accumulation in GPU server
  6. Firmware or driver anomalies
  7. Hardware degradation

L1 Engineer Actions

Step 1 - Log the Alert

  1. Log in to the monitoring platform
  2. Review:
    1. Alert summary
    2. Affected node details

Step2 - Create Ticket

  1. Create an internal incident ticket including:
    1. Alert name
    2. Node hostname
    3. Timestamp
    4. GPU ID (if available)
    5. Screenshot or alert logs

Step 3 – Initial Verification

  1. Check whether the alert is:
    1. Transient OR
    2. Persistent

Step 4 – Escalation (If required)

  1. If alert persists beyond threshold:
    1. Escalate to L2 support

Step 5 – Documentation

  1. After confirmation with L2:
    1. Update Issues Tracker Sheet
    2. Add troubleshooting steps and observations in ticket

L2 Engineer Actions

Step 1 – Check GPU Temperature & Utilization

nvidia-smi
Verify:
  1. GPU memory temperature
  2. GPU utilization
  3. Running GPU processes
  4. Power consumption

Step 2 – Check System Memory Status

free -h

Verify:
  1. Available memory
  2. Memory pressure

Step 3 – Identify Running GPU Processes

top OR htop
ps -aux

Check for:
  1. Long-running jobs
  2. Unexpected GPU usage
  3. High CPU/ GPU consumers

Step 4 – Check Zombie Processes

ps aux | grep Z

  1. Identify zombie processes
  2. Investigate parent processes

Step 5 – Verify Server Airflow & Cooling

  1. Physically check:
    1. Server fans are operational
    2. No airflow obstruction
    3. Rack cooling is functioning
    4. No dust accumulation
  2. If issue found:
    1. Inform data center operations team

Step 6 – Verify System Memory Configuration

cat /proc/meminfo

Step 7 – Verify Installed Memory Hardware

sudo dmidecode -t memory

Confirm:
  1. Correct memory modules
  2. No hardware mismatch
  3. All DIMMs detected

Step 8 – OEM Escalation (If required)

  1. Escalate if:
    1. GPU memory temperature remains critical
    2. Hardware anomaly detected
    3. Cooling components malfunction
    4. Repeated thermal alerts
  2. Provide:
    1. nvidia-smi output
    2. System logs
    3. Hardware configuration
    4. Sever serial number

Resolution

The issue is considered resolved when:
  1. GPU memory temperature returns to normal range
  2. No recurring alerts in monitoring system
  3. Workloads run without thermal throttling

Escalation

  1. L1 → L2: If alert persists beyond threshold
  2. L2 → OEM/Vendor: If hardware or thermal issue persists

    • Related Articles

    • GPU Temperature Critical alert - Troubleshooting Guide

      Purpose This article provides guidance for engineers to identify and respond to the GPU Temperature Critical alert in GPU compute nodes. It outlines the alert meaning, possible causes, and the required L1 and L2 troubleshooting steps. Alert Name GPU ...

    • 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 ...

    • 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 - ...

    • GPU Temperature Warning alert- Troubleshooting Guide

      Purpose Provide guidance for engineers to investigate and resolve the GPU Temperature Warning alert in GPU compute nodes. When To Use Use this article when the monitoring system generates the alert: GPU Temperature Warning Alert Details This alert is ...

    • CPU Temperature Critical - Troubleshooting Guide

      Purpose Provide a standard troubleshooting procedure for engineers when a CPU Temperature Critical alert is received in the cluster. This article helps engineers quickly identify the cause and perform initial remediation before escalation. When to ...