Thank you for your initial guidance. I wanted to share the resolution steps I took, as this detailed feedback may be beneficial for others in the community, particularly those using the Yahboom carrier board.
Initial Resolution (Starting from Scratch)
To address the initial issue, I performed a clean flash:
I used the Yahboom factory image to overwrite the SSD.
I ensured the Yahboom “super boot” was correctly reimplemented. This allowed me to start with a fresh, error-free operating environment.
Subsequent Issue and the Core Fix
When attempting the system updates on the newly flashed OS, I encountered a new error during the sudo apt upgrade process:
Error Encountered:
Processing triggers for nvidia-l4t-initrd (36.4.4-20250616085344) ...
/usr/sbin/nv-update-initrd: line 127: cpio: command not found
Cleaning up the temporary directory for updating the initrd..
dpkg: error processing package nvidia-l4t-initrd (--configure):
installed nvidia-l4t-initrd package post-installation script subprocess returned error exit status 127
Errors were encountered while processing:
nvidia-l4t-initrd
E: Sub-process /usr/bin/dpkg returned an error code (1)
The root cause was the missing standard Linux utility, cpio. The post-installation script for nvidia-l4t-initrd requires cpio to rebuild the initial ramdisk (initrd).
Solution Steps
I resolved the cpio: command not found error by manually installing the missing package:
Install the missing dependency (cpio):
sudo apt install cpio
Ensure all pending configurations were completed: (The installation of cpio triggered the successful configuration of nvidia-l4t-initrd, but this step confirms it.)
sudo dpkg --configure -a
Finalize the Update: After confirming no errors remained, I performed the reboot. The system successfully applied the remaining updates and booted without issue.
While my initial setup involved a specialized carrier board, the cpio dependency error is a common software issue that could occur on any Jetson configuration. I hope this detailed breakdown helps others who encounter a similar dependency failure during a JetPack update.
Possible reasons here are the auto update didn’t update successfully. Partially update makes the kernel image goes to new version while kernel module drivers are still the old one. This would make PCIe driver, which is a kernel module, fails to get loaded and lead to NVMe not able to be found.
Also, another possible reason is your custom board is using custom BSP since beginning and auto-upgrade will install NV driver to make your custom BSP gone.
Thanks for explaining some potential reasons for this issue, sounds plausible, as it failed to load the NMVe due to an incorrect PARTUUID.
We are currently moving from a PoC to an MVP; do you recommend any specific vender that sells pre-flashed Jetson devices or do you have a partner list?
No matter what partner you get your carrier board. The key point here if you want to run apt-get upgrade, then you need to remove NVIDIA related items from your apt source list before doing that.
NVIDIA apt source will install NVIDIA default driver again but a custom carrier board should never totally rely on NVIDIA default driver. It may crash functionality of a custom board.
Also, should not believe any pre-flashed thing. You should always learn how to totally reflash a board.
Thanks for explaining the key point, that I did not know (an unknown unknown).
Thanks for giving a heads-up that we should always be able to completely reflash a board, we will make sure that is the case, and thanks for referencing a partner list.