Setting nvidia runtime in docker on CentOS 7.5 is failing

[skshelly]

On CentOS after installing docker 19, Nvidia-docker v2 & CUDA the following command is failing.

• ExecStart=/usr/bin/dockerd -H fd:// --default-runtime=nvidia --node-generic-resource gpu=${GPU_ID}

• The GPU_ID is obtained from nvidia-smi -L

The same combination of commands succeeds on Ubuntu 18:04.

My goal is to set up the above such that I can use GPUs when running docker-containers in the swarm mode - on CentOS 7.5

Any insights into this appreciated.