The workaround is to restart the
nv-hostengine service on the affected ESXi host.
Here are the steps:
- Login to the affected ESXi host as root
- Run the below command to stop the service: nv-hostengine -t
- Run the below command to start the service: nv-hostengine -d
- ps | grep nv-hostengine (lists the running nv-hostengine processes)
Once the service is started, wait for at least 15 minutes for the GPU data to reflect in the dashboards.