`./scripts/docker_build_jetpack.sh` fail

[SkalskiP]

Hi, when I'm running ./scripts/docker_build_jetpack.sh on my Jetson Xavier it fails during the execution of this part of Dockerfile.jetpack:

#
# install CUDA Toolkit
#    
RUN apt-get update && \
    apt-get install -y --no-install-recommends \
		  cuda-toolkit-* \
    && rm -rf /var/lib/apt/lists/* \
    && apt-get clean

The error message that I'm getting is:

Unpacking cuda-nvrtc-10-2 (10.2.89-1) ...
dpkg: error processing archive /tmp/apt-dpkg-install-HjrbFB/62-cuda-nvrtc-10-2_10.2.89-1_arm64.deb (--unpack):
 unable to create '/usr/local/cuda-10.2/targets/aarch64-linux/lib/libnvrtc.so.10.2.89.dpkg-new' (while processing './usr/local/cuda-10.2/targets/aarch64-linux/lib/libnvrtc.so.10.2.89'): Read-only file system
dpkg: error while cleaning up:
 unable to remove newly-extracted version of '/usr/local/cuda-10.2/targets/aarch64-linux/lib/libnvrtc.so.10.2.89': Read-only file system
dpkg-deb: error: paste subprocess was killed by signal (Broken pipe)
Selecting previously unselected package cuda-nvrtc-dev-10-2.
Preparing to unpack .../63-cuda-nvrtc-dev-10-2_10.2.89-1_arm64.deb ...
Unpacking cuda-nvrtc-dev-10-2 (10.2.89-1) ...
dpkg: error processing archive /tmp/apt-dpkg-install-HjrbFB/63-cuda-nvrtc-dev-10-2_10.2.89-1_arm64.deb (--unpack):
 unable to create '/usr/local/cuda-10.2/targets/aarch64-linux/include/nvrtc.h.dpkg-new' (while processing './usr/local/cuda-10.2/targets/aarch64-linux/include/nvrtc.h'): Read-only file system
dpkg: error while cleaning up:
 unable to remove newly-extracted version of '/usr/local/cuda-10.2/targets/aarch64-linux/include/nvrtc.h': Read-only file system
...

@dusty-nv or anybody else, do you maybe happen to know why is that?

[dusty-nv]

Hi, you shouldn't need this anymore, instead please use l4t-jetpack container from NGC. I should remove it from this repo.

---

From: Piotr Skalski @.> Sent: Tuesday, August 30, 2022 2:00:56 PM To: dusty-nv/jetson-containers @.> Cc: Dustin Franklin @.>; Mention @.> Subject: [dusty-nv/jetson-containers] ./scripts/docker_build_jetpack.sh fail (Issue #190)

Hi, when I'm running ./scripts/docker_build_jetpack.sh on my Jetson Xavier it fails during the execution of this part of Dockerfile.jetpack:

install CUDA Toolkit

RUN apt-get update &&
apt-get install -y --no-install-recommends
cuda-toolkit-*
&& rm -rf /var/lib/apt/lists/*
&& apt-get clean

The error message that I'm getting is:

Unpacking cuda-nvrtc-10-2 (10.2.89-1) ... dpkg: error processing archive /tmp/apt-dpkg-install-HjrbFB/62-cuda-nvrtc-10-2_10.2.89-1_arm64.deb (--unpack): unable to create '/usr/local/cuda-10.2/targets/aarch64-linux/lib/libnvrtc.so.10.2.89.dpkg-new' (while processing './usr/local/cuda-10.2/targets/aarch64-linux/lib/libnvrtc.so.10.2.89'): Read-only file system dpkg: error while cleaning up: unable to remove newly-extracted version of '/usr/local/cuda-10.2/targets/aarch64-linux/lib/libnvrtc.so.10.2.89': Read-only file system dpkg-deb: error: paste subprocess was killed by signal (Broken pipe) Selecting previously unselected package cuda-nvrtc-dev-10-2. Preparing to unpack .../63-cuda-nvrtc-dev-10-2_10.2.89-1_arm64.deb ... Unpacking cuda-nvrtc-dev-10-2 (10.2.89-1) ... dpkg: error processing archive /tmp/apt-dpkg-install-HjrbFB/63-cuda-nvrtc-dev-10-2_10.2.89-1_arm64.deb (--unpack): unable to create '/usr/local/cuda-10.2/targets/aarch64-linux/include/nvrtc.h.dpkg-new' (while processing './usr/local/cuda-10.2/targets/aarch64-linux/include/nvrtc.h'): Read-only file system dpkg: error while cleaning up: unable to remove newly-extracted version of '/usr/local/cuda-10.2/targets/aarch64-linux/include/nvrtc.h': Read-only file system ...

@dusty-nvhttps://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fdusty-nv&data=05%7C01%7Cdustinf%40nvidia.com%7C937b9411e20442fe1d3608da8ab19691%7C43083d15727340c1b7db39efd9ccc17a%7C0%7C0%7C637974792587162012%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=vtft5dK2ZoCnk1XTVGFZASvLIwsOmLlIlPrfgt%2Bxyr4%3D&reserved=0 or anybody else, do you maybe happen to know why is that?

— Reply to this email directly, view it on GitHubhttps://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fdusty-nv%2Fjetson-containers%2Fissues%2F190&data=05%7C01%7Cdustinf%40nvidia.com%7C937b9411e20442fe1d3608da8ab19691%7C43083d15727340c1b7db39efd9ccc17a%7C0%7C0%7C637974792587318256%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=Fmmbb29U3EUO9D2iEiy4csXHKJ%2B2%2FN7AxNxVuj1zo4A%3D&reserved=0, or unsubscribehttps://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FADVEGK5B2V6FZOYTTTCDOGDV3ZD5RANCNFSM6AAAAAAQAUPLDI&data=05%7C01%7Cdustinf%40nvidia.com%7C937b9411e20442fe1d3608da8ab19691%7C43083d15727340c1b7db39efd9ccc17a%7C0%7C0%7C637974792587318256%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=aDxk4TSP9utde9g5IFa0%2FCAZl6R2MmdBZ2GA%2BRoZJgY%3D&reserved=0. You are receiving this because you were mentioned.Message ID: @.***>

[SkalskiP]

@dusty-nv do you mean this container: https://catalog.ngc.nvidia.com/orgs/nvidia/containers/l4t-jetpack? It only supports JetPack 5.x, at least judging by available tags. I'm looking for a base image containing TensorRT that I could use with JetPack 4.5.0.

[dusty-nv]

I'm looking for a base image containing TensorRT that I could use with JetPack 4.5.0.

You should use l4t-base container with --runtime nvidia and TensorRT will automatically be mounted in from the host device. The "fat containers" which have CUDA/cuDNN/TensorRT installed inside the containers themselves only started with JetPack 5.x

BTW I've removed the jetpack.dockerfile from jetson-containers repo in commit https://github.com/dusty-nv/jetson-containers/commit/39496f8eba51ababb0cfb625fc70410163e4fe43

[SkalskiP]

Okay, I'll definitely try. I have a question, however. My goal is to make TensorRT accessible through python 3.8 installed in virtual env inside docker image. IS that possible?

[dusty-nv]

I think you would have to rebuilt the TensorRT Python bindings for Python 3.8 - i haven't tried this. I think they are on the TensorRT github though

[SkalskiP]

Is that what you mean https://github.com/NVIDIA/TensorRT/tree/main/python?

[dusty-nv]

Yes, I believe that's it

[SkalskiP]

Thanks I'll give it a try!