Open austinmw opened 2 years ago
Having the same issue
Thank you for reporting the issue! Please let us know if you are still facing the issue.
Having the same issue with images with torch<=1.9, any updates on how to mitigate?
You can give this a try in the docker file before apt-get update
# Workaround for CUDA Linux Repository Key Rotation
# https://forums.developer.nvidia.com/t/notice-cuda-linux-repository-key-rotation/212772
RUN rm /etc/apt/sources.list.d/cuda.list
RUN rm /etc/apt/sources.list.d/nvidia-ml.list
RUN apt-key del 7fa2af80
RUN apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/3bf863cc.pub
RUN apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1804/x86_64/7fa2af80.pub
Checklist
Concise Description: Unable to
apt update
SageMaker DLC's due to failing NVIDIA certificate verificationTo reproduce:
nvidia-docker run -it --rm 763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-inference:1.10.2-gpu-py38-cu113-ubuntu20.04-sagemaker "apt update"
DLC image/dockerfile: Multiple, for example:
763104351884.dkr.ecr.us-west-2.amazonaws.com/pytorch-inference:1.10.2-gpu-py38-cu113-ubuntu20.04-sagemaker
Current behavior:
Expected behavior: Completing successfully
Additional context: This largely prevents using these images as a base to build on top of