Install driver in Dockerfile on TX2

I’m having issues trying to get the TX2 drivers installed into my docker image.

This is the error i keep running up against. Any help is greatly appreciated

Step 8/24 : RUN chown root /etc/passwd /etc/sudoers /usr/lib/sudo/ /etc/sudoers.d/README
—> Running in a4b38714f2f0
Removing intermediate container a4b38714f2f0
—> d943b42b1cf8
Step 9/24 : RUN /tmp/Linux_for_Tegra/ -r / && rm -fr /tmp/*
—> Running in 3a57be800519
Using rootfs directory of: /
Extracting the NVIDIA user space components to /
Extracting the BSP test tools to /
Extracting the NVIDIA gst test applications to /
Extracting Weston to /
Extracting the configuration files for the supplied root filesystem to /
tar: etc/hostname: Cannot open: File exists
tar: etc/hosts: Cannot open: File exists
tar: Exiting with failure status due to previous errors
The command ‘/bin/sh -c /tmp/Linux_for_Tegra/ -r / && rm -fr /tmp/*’ returned a non-zero code: 2


What kind of driver do you indicate here?
Do you mean the GPU driver?

Please noticed that Jetson’s GPU driver is integrated into the OS image. It is not a standalone process.
We have a container called l4t-base:r32.2 based on JetPack 4.2.1 environment. It may just meet your requirement:


I will give that a go. I am trying to install tensorflow 1.14 with GPU and Jupyter with GPU and have not had much luck


Just want some clarification first.

Do you want to install TensorFlow on the Jetson or on the docker image running on the Jetson?
For the former, you can follow this document:

For the latter, you will need to install all the required packages first.
Our container only includes L4T and CUDA.


That is actulally where I started from and those instructions do not work in a docker container. I have got it figured out now after some bashing around. Thank you for your input.

I will attach my solution after the TX2 finishes running the Dockerfile