Lost one GPU in nvidia-smi

I have a DGX server with 8*H100/NVLINK. Now it losts one GPU with nvidia-smi. I tried to restart the node, the issue is not recovered. I attach the log below. Any suggestions?

nvidia-bug-report.log.100.64.24.63.gz (12.1 MB)

Hello,

For DGX support, please see this: