SDK Manager 1.6 will not install on debian11

Trying to install SDK Manager 1.6.1.8175 on the host machine running debian11 distribution.
It seems that the installer does not like the OS version.

Is there a way to force the SDK Manager installation?


Update:
I spoofed the system ids by changing /etc/os-release temporarily with

NAME=“Ubuntu”
VERSION=“18.04”
ID=ubuntu
ID_LIKE=debian
PRETTY_NAME=“Ubuntu 18.04”
VERSION_ID=“18.04”
HOME_URL=“https://www.ubuntu.com/
SUPPORT_URL=“https://help.ubuntu.com/
BUG_REPORT_URL=“https://bugs.launchpad.net/ubuntu/
PRIVACY_POLICY_URL=“Data privacy | Ubuntu
VERSION_CODENAME=focal
UBUNTU_CODENAME=focal

The SDK installer is now downloading. Will update the thread after installation with the status.

Any result to share?

I had to add


wget -qO - https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/7fa2af80.pub | sudo apt-key add -

I also had to download and install a whole lot of packages that the downloader wouldn’t download from (Index) - I guess this would be repo issues.

The system locale needed to be rebuild to allow for the switch to US.UTF-8 (I’m in Australia). If this is not done, the dpkg-query fails.

Unfortunately, the VPI fails due to the Python2-numpy package missing

dpkg: dependency problems prevent configuration of python-vpi1:
python-vpi1 depends on python2.7-numpy; however:
Package python2.7-numpy is not installed.

I couldn’t find the actual package online, but I did instal the numpy for p2.7. Unfortunately it doesn’t work still, and if I install the python-vpi1 ignoring the dependencies then the apt reports it and the whole process stops.

By checking with internal team, the SDK Manager can’t work well on debian11 distribution, the Ubuntu18.04 is suggested and suitable for Jetson device as well.

Thank you for checking with the team.

I got a bit more progress, found the python2 package and am ploughing through with packages for Python3 now.

I understand the complexities behind the system like this, however, the ubuntu and debian are very close.
I’ll try to fix this before I decide if I’m happy to throw away my OS just to use the Xavier.

I’ve managed to install the SDK.

I had to install python3.6 from a separate source and install numpy via pip, not package.

Then CUDA package was installed manually, overriding dependency on python3.6-numpy and the dependency was removed from the cuda package in the /var/lib/dpkg/status.

After that, everything was completed on its own.
I haven’t tried if everything actually runs but it’s a start.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.