#Release Notes

#v0.18.2

#Features

• Bump NVIDIA Container Toolkit to v1.18.2

#Fixed Issues

• Watch and hot reload the updated certificate for webhook server
• Ensure that cdi.FeatureFlags are passed to CDI library
• Fix race condition in config-manager when label is unset
• Fix nested container use cases by ensuring that IPC sockets are not mounted readonly

#v0.17.4

#What's Changed

• Auto set runtimeClass for gpu pod and not need to modify default runtime manually
• Bump golang from 1.22.2 to 1.24.7
• Bump github.com/NVIDIA/nvidia-container-toolkit from 1.17.6 to 1.17.8
• Bump github.com/NVIDIA/go-nvml from 0.12.4-1 to 0.12.9-0
• Bump github.com/NVIDIA/go-nvlib from 0.7.1 to 0.7.4
• Ensure that directory volumes have Directory type
• Ignore errors getting device memory using NVML

#v0.17.2

#New and Optimized Features

This is an independent plug-in of the NVIDIA official Device Plugin, providing complete physical GPU isolation and NUMA-aware scheduling. Its functional positioning is consistent with the pGPU Module in the original GPU manager.

#Fixed Issues

No issues in this release.

#Known Issues

No issues in this release.