Resetting Jeston TX2 to Out Of Box State

Hello,

I received a NVIDIA Jetson TX2 and when I powered on the board the user was set as admin instead of nvidia/ubuntu like it is referenced in the manual. The team member I received the board from didn’t create the user and I haven’t seen anything in the manual about it. Is there a way I can reset the OS to out of the box state so that I can log in using the nvidia credentials.

Thank you

Just so you are aware, the earlier installs created login name and pass by default. Later releases (due to California law) stopped providing a default user name and pass. Whether or not you have this will depend on the L4T release shipped.

In later releases there is a “first boot” setup for timezone and adding your user name/pass.

It is likely you’d be advised to start by flashing the most recent release anyway, but what you see may not be an error. Older documentation will only be valid for older L4T releases. If you were to flash with a new install, then you’d still need to create the user and password…if you choose user name “nvidia”, then that is fine, but any name you choose would work, and login as “nvidia” would fail unless you actually created that account.

Do you have serial console available? If so, can you check if this is offering to allow you to perform a first boot setup? Serial console info here:
http://www.jetsonhacks.com/2017/03/24/serial-console-nvidia-jetson-tx2/

Is this the guide you would follow to flash the most recent release? Welcome — Jetson Linux<br/>Developer Guide 34.1 documentation

I’m not sure what the best way to go about it is. I’ve reflashed some OS before using a SD card or Flash drive but I’m not sure if that would apply to the Jetson TX2.

I don’t know if the document you mentioned is the most recent, but it does include first boot setup and the oem-setup script, so this document will work and is sufficiently recent. The best documentation is always the one released for what you have currently flashed. The URLs for a list of L4T and JetPack/SDKM content may be useful to you (you might need to go there, log in, and then go there a second time if redirect fails):
L4T Listings
JetPack/SDKM Listings

Normally a TX2 would be flashed for eMMC and not use an SD card. If flashing on command line, then a TX2 dev kit would be something like this from the “Linux_for_Tegra/” subdirectory:
sudo ./flash.sh jetson-tx2 mmcblk0p1
(if you’ve flashed once before with JetPack/SDKM, then you will have directory “~/nvidia/nvidia_sdk/JetPack...version.../Linux_for_Tegra/”, and the “rootfs/” subdirectory to this will already be populated)

The above command line is essentially what JetPack/SDKM would run behind the scenes if using the GUI to flash. Assuming this is not a custom carrier board then nothing else is needed; if a custom carrier board is involved, then you’d basically need the manufacturer’s board support package (which in turn is basically an overlay of content on top of the “Linux_for_Tegra/” content).

1 Like

Great this is the approach I have been trying to take but I have been running into an issue referenced here: Jetson Tx2 Probing Target Board Failed

Do you know of any solutions/things I may be doing wrong?

Thank you

When probing target board fail messages occur there is often something wrong in the flash setup, and is a different question than the actual flash. Sometimes this is due to using a VM, sometimes it is due to using a charger cable instead of a quality cable, sometimes it is due to using the wrong specification in the flash command for a custom board. I couldn’t answer for your particular case, but maybe it is because of using an Ubuntu 20.04 host (you should use 18.04, but people have made other releases work…but there is no guarantee).

The flash software itself won’t care if you are using Ubuntu of some odd release, or Fedora, or any other flavor of Linux. It is the GUI JetPack/SDK Manager which cares (the GUI does some downloading, but other than passing commands to other software, has nothing to do with the actual flash program). So flashing on command line from Ubuntu 20.04 should “just work”, but only if your cable is correct, your board specification is correct, so on.