If nvidia-docker ps -a does not respond,
admin@server:~$ sudo service docker restart
admin@server:~$ sudo docker ps - -all
the docker daemon service might be dead.
To display DGX topology:
nvidia-smi topo -m
To query device:
/usr/local/cuda/samples/1_Utilities/deviceQuery/deviceQuery
To raise tickets for any DGX related problems:
IT support for DGX-1(send to both):