Error: "Could not initialize plugin '/usr/lib64/vmware/plugin/libnvidia-vgx.so'" when powering on a VM using an NVIDIA A16 GPU

Products

• VMware vSphere ESXi

Issue/Introduction

When attempting to power on a virtual machine, you encounter the following error message preventing the VM from starting:

Could not initialize plugin '/usr/lib64/vmware/plugin/libnvidia-vgx.so' for vGPU 

Additionally, when checking the host via the nvidia-smi command, you observe a FAN ERR! . This error hampers the ability of the virtual machine to utilize the vGPU card.

Environment

 

• VMware ESXi / vSphere (e.g., ESXi 8.0)

• NVIDIA A16 GPU

• NVIDIA GRID VIBs

 

Cause

 "FAN ERR!" in the nvidia-smi output,  can  lead to   the virtual machine to detect the vGPU as corrupted during the power-on process.

Resolution

To properly address this issue , it is highly recommended that you engage your hardware vendor (e.g., HP or NVIDIA) for further diagnostics.

Additional Information

Workaround

To troubleshoot and attempt to temporarily clear the sensor error condition, you can perform the following steps:

• Reboot the Host: Perform a full power cycle of the physical ESXi server. This action resets the sensor and may clear the error condition.

• Check Host Drivers: Ensure that your ESXi VIB/driver and NVIDIA GRID VIBs are fully updated and compatible with your environment.

• Update Firmware: Verify that your server’s management controller (such as iLO, iDRAC, or XClarity) runs the latest vendor-approved firmware and does not cause firmware conflicts.