rongfu.leng

We can remove cert-manager this depend

8

Current `harbor-operator` depends on too many components, we hope can remove `cert-manager` this depend.

helm dependency can wait other helm release install success, again install next helm release

1

Output of `helm version`: Output of `kubectl version`: Cloud Provider/Platform (AKS, GKE, Minikube etc.):

Use NVIDIADriver CRD to install GPU driver in a centos7 and a ubuntu22.04

1

I currently have two nodes, one with Ubuntu 22.04 operating system and one with CentOS 7 operating system. I use `NVIDIADriver` for heterogeneous installation of drivers, but I found that...

add id mount

2

Fixes: https://github.com/containers/youki/issues/2307

change `mount.rs` find_parent_mount from pub to private

4

The current `find_parent_mount` method is only in `impl Mount` use, we can change to `private fn` from `pub fn`. https://github.com/containers/youki/blob/4d4ce7c3710d1ed8f155adf23fb62dc672197043/crates/libcontainer/src/rootfs/mount.rs#L614

What can we help you? Current harbor still stopping 2.6.x version ,but https://github.com/goharbor/harbor/security/advisories/GHSA-mq6f-5xh5-hgcf this vulnerability require min 2.7.3 version.

add warning when set memory limits check is pre-volta arch

2

Fixes: https://github.com/NVIDIA/k8s-device-plugin/issues/647#issuecomment-2124122930

add validator nvidia driver whether install success

13

Fixes: https://github.com/Project-HAMi/HAMi/issues/136 Add a `driver-validator` Daemonset workload, in the node to valid Nvidia driver when the install is a success. if install success, add node label `hami.io/driver-validator="true"` else `hami.io/driver-validator="false"`, `device-plugin`...

if use gpu-operator install nvidia driver, when node restart, driver-plugin start CrashLoopBackOff

3

1. when the node reboots after, `gpu-operator` can reinstall NVIDIA `driver`; 2. but the same time `device-plugin` pod starts, but this `driver` does not install completely. so pod in CrashLoopBackOff....