NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver on RHEL 8.3
Nov 7 2020 - 1 min read
On RHEL 8.3:
$ cat /etc/redhat-release
Red Hat Enterprise Linux release 8.3 (Ootpa)
If you hit this error after the CUDA driver installation:
$ nvidia-smi
NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.
You can blacklist the “nouveau” driver at the end of the line “default_kernelopts” of /etc/grub2.cfg add: “ rd.driver.blacklist=nouveau modprobe.blacklist=nouveau “.
Check the status after your change:
$ sudo grep default_kernelopts /etc/grub2.cfg
set default_kernelopts="root=/dev/mapper/vg0-LVroot ro nofb quiet splash=quiet console=tty0 crashkernel=auto resume=/dev/mapper/vg0-LVswap rd.lvm.lv=vg0/LVroot rd.lvm.lv=vg0/LVswap biosdevname=0 console=ttyS1,115200n8 rd.driver.blacklist=nouveau modprobe.blacklist=nouveau "
Apply the changes:
$ sudo grub2-mkconfig -o /boot/grub2/grub.cfg
Reboot:
$ sudo systemctl reboot