Update AWS EKS documentation to use an Ubuntu AMI #195
Labels
bug
Something isn't working
cloud/aws
Amazon Web Service cloud
platform/kubernetes
Runs on Kubernetes
Thank you for the phenomenal AWS documentation your team maintains,
eksctl
currently uses the "Amazon Linux 2 x86 Accelerated AMI" by default which has GPU driver version 470.161.03.As of this week the NVIDIA GPU Operator officially supports EKS for Ubuntu AMIs in release 23.3.0.
Using the GPU Operator with the current default AMI results in the driver container not being deployed due to the pre-installed drivers and the device-plugin-validator fails likely due to the old GPU drivers in the cluster.
I recommend we wait to update the documentation until this issue is resolved so that we can provide a really clean way for users to create a managed Ubuntu nodegroup: eksctl-io/eksctl#6499
Once this is implemented the only change needed in RAPIDS documentation is changing the existing
eksctl cluster create
command to include the additional flag:This should provide users with the latest recommended GPU drivers and resolve the device plugin validator pod issue.
The text was updated successfully, but these errors were encountered: