REQUEST: Can someone help me debug and resolve this issue?
I was performing my first ever NVIDIA system setup with a Jetson Orin NX (16GB) dev kit board with a M.2 2280 NVMe drive and an Ubuntu 20.04 LTS virtual machine (~3.5GB RAM, 40GB disk) running from a Windows 10 machine. I saw the warning that 8GB of RAM was needed for performance but I continued anyways.
Unfortunately the flash failed at 15.87% (see screenshot from SDK Manager below) and the SDK Manager went fully unresponsive so I could not get the logs at the time of incident but when I power cycled the VM I could export some of the logs (see below zip file). When I tried running sdkmanager once more, I got a ENOSPC error on my VM’s shell.
From there, I powered off the Orin NX board and called it a night but the next day (as I am writing this) I tried again on an Ubuntu 20.04 native host laptop from a colleague of mine and now I cannot begin the flash process due to an alleged issue with USB (even though this same setup had no issues a day earlier and I tried multiple USB-A to USB-C cables per the instructions). Note that I tried replacing the NVMe drive with a new one in case it was corrupted and I get the same results which makes me believe something may be wrong on the Orin module itself.
@kayccc I did use a native Ubuntu 22.04 laptop (System76) for the reflash, which is where I am now having problems. I now know that VM is not supported but is there a way to recover my Orin NX module or is it bricked?
Please put the Orin NX in recovery mode and connect flashing cable with your Ubuntu 22.04 laptop.
And execute lsusb command to check whether the device in recovery mode.
I think the USB is working (and it was before, as well, since SDKM recognized the Orin NX that was plugged in) but I still think there could be something wrong from when the flash failed at ~20% progress. Note that I plugged in the Orin NX over USB, ran lsusb, unplugged it, ran lsusb, and then retried to verify the operation.
I tired 2 things this morning on my native Ubuntu 20.04 host laptop:
Put Orin NX into recovery mode (shorting the GND/FC_REC pins with a jumper wire) and holding it in recovery mode by not removing the wire during the flash attempt
Put Orin NX into recovery mode (shorting the GND/FC_REC pins with a jumper wire) and releasing the FC_REC pin after about 30sec and then attempting to flash
In both attempts, I get the same USB issue and I can assure you the cable is a good one and I can see the NVIDIA device on the lsusbshell dialogue.
Both of the aforementioned attempts should be covered by the following debug log:
I still have a feeling that something is corrupted or locked out due to the previous failed flash attempt while working from the Ubuntu VM. Please let me know your thoughts from the above logs when you get a chance.
Hey @DavidDDD@kayccc any chance you have ideas for how to restore the function of my Orin NX unit given the provided logs and information? I am hoping it is not permanently bricked and I am also hoping to get it back to functionality ASAP but I have not found any documentation (yet) that would help me debug or find a fix in my current scenario. Is there a way to reset the module beyond the force recovery option to undo the partial flash I started?
I have also seen @linuxdev comment on a lot of these issues in the past, any chance you would be able to help in this case?
Hi,
From your source list. there are duplicate package for https://download.docker.com/linux/ubuntu
deb [arch=amd64] https://download.docker.com/linux/ubuntu focal stable
# deb-src [arch=amd64] https://download.docker.com/linux/ubuntu focal stable
deb https://deb.nodesource.com/node_14.x focal main
# deb-src https://deb.nodesource.com/node_14.x focal main
deb http://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/ /
# deb-src http://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/ /
deb https://repos.influxdata.com/debian stable main
# deb-src https://repos.influxdata.com/debian stable main
deb https://download.docker.com/linux/ubuntu focal stable
# deb-src https://download.docker.com/linux/ubuntu focal stable
Please remove one of package and rerun the sdk manager
In the screenshot I saw “no space left on device”. I don’t know if that applied to the Jetson or to the host PC. On your host PC, if you first cd to the location of the flash software, e.g., somewhere at: ~/nvidia/nvidia_sdk/JetPack...version.../Linux_for_Tegra/
…what do you see from this command: df -H -T .
(if that was an old “not enough space” message, still check this out even on the new platform because typically you need a lot more than 40 GB of disk on the host PC side just for the temp flash content; then there is the actual flash software on top of that)
Assuming you are still at ~/nvidia/nvidia_sdk/JetPack...version.../Linux_for_Tegra/ there will be a subdirectory, “bootloader/”. What do you see from: ls -l system.img*
You said you were flashing to an m.2 NVMe. What is the size of the partition you are using?
Also, are you using a dev kit from NVIDIA, or is the carrier board from a third party manufacturer? The dev kit carrier board is easy to deal with, third party carrier boards require patching or different software from the third party manufacturer if the carrier board differs in any way from the dev kit carrier board.
Btw, it is nearly impossible to “brick” a Jetson. They don’t have a BIOS. All of the equivalent of a BIOS is part of the flash in ordinary software.
