3D Virtual Machine fails to power on

|
  • 6
  • 0

Issue Description

When the 3D virtual machine is turned on, it prompts that the GPU is in shortage and no GPU can be assigned to T4-2Q.

Error/Warning Information

Handling Process

1.The graphics card can be recognized normally in the background.
2.Check the nvidia-vgpu-mgr.log under the Nvidia service key log /sf/log/today/ directory, prompting an ECC error.
3. Execute /sf/data/local/sgax/sgax-chroot.sh to enter the graphics card mode, and then execute nvidia-smi -q | grep -i -A2 "Ecc Mode" to check whether the ecc of all the graphics cards is closed, and it is found that the graphics card ECC is enable state, it needs to be closed.
4.Execute nvidia-smi -e 0 to turn off the ECC of the graphics card and restart the host. After restarting the host, check that the status of the graphics card is disabled, and then boot normally.

Root Cause

1.The graphics card in the vGPU environment currently does not support enabling ECC. It is normal to disable ECC.

Solution

1.Turn off ECC in the backend.

I want to write a case
Doc ID: 3908
Author: Sangfor_Siva
Updated: 2024-11-12 16:41
Version: