Skip to content

[Bug] a security change in version 122 breaks GPU node setup #6025

@wjxiz1992

Description

@wjxiz1992

What were you trying to accomplish?

setup an eks cluster with GPU node

What happened?

nvidia-device-plugin ds is broken:

Error: container has runAsNonRoot and image will run as root (pod: "nvidia-device-plugin-daemonset-dc78h_kube-system(eee1d2bb-3b7d-44ba-86c6-610413f56a6c)", container: nvidia-device-plugin-ctr)

How to reproduce it?

create a cluster with gpu node:

...
--node-type g4dn.8xlarge
...

Versions
0.122

This should introduced by #5970 where you add a security limit for user type. When I use an older version, I cannot reproduce this error.

Metadata

Metadata

Assignees

Labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions