Jetson-Nano-Ubuntu-20-image
Jetson-Nano-Ubuntu-20-image copied to clipboard
Minimum SD card requirements and image validation
Hello! Just wanted to follow up from https://github.com/Qengineering/Jetson-Nano-Ubuntu-20-image/issues/92. The JetsonNanoUb20_bare.img.xz image downloaded successfully and appeared to be written to a known-good 64GB SD card using Etcher running on Pop!_OS 22.04 LTS, but it failed validation twice. While the .xz was ~5.6GB, Etcher indicated 32GB for the install. As the Installation section of the README shows a 32GB card would suffice for the JetsonNanoUb20_3b.img.xz image, I wanted to recommend two things:
- My understanding was that the
..._bareimage should not include the additional 21GB of software, as outlined here. Since the..._bareimage was more than 32GB installed, it might be worth validating the integrity of each image. - If the images are intact, kindly update the minimum storage requirements for an SD card compatible with either image.
I wound up using the following image from NVIDIA: https://developer.nvidia.com/embedded/downloads#?search=4.6.1&tx=$product,jetson_nano
I manually upgraded to Ubuntu 20.04 LTS, but this came with its own set of headaches around libopencv and chromium, similar to the caveats outlined in Qengineering's instructions at https://qengineering.eu/install-ubuntu-20.04-on-jetson-nano.html.
I hope this helps others who might be in a similar situation with needing a newer OS for the Jetson Nano.
Most flashing problems are related to the size of the SD card used. If a card has only a few bytes less than the one I used, Etcher fails. As you know, all manufacturers produce cards with slightly different sizes, even if they are all about 32 GB. They usually differ by a few bytes. I tried to use the 'smallest' card, but there are undoubtedly cards on the market that are just a little bit smaller.
If you use a 64 GB card, you will avoid all these problems. Afterwards, enlarge the partition with GParted, and you will have enough working space for your projects.
Thank you for the reply. I apologize, I might not have been clear in my phrasing. As it turns out, I was using a 64GB card. I thought it was strange the ..._base image failed to verify as the card was freshly formatted to FAT32 and had previously been running Ubuntu for a Raspberry Pi 5. I reformatted the card after the ..._base image failures and was able to write the image from NVIDIA with no issues, suggesting the problem is unlikely to be the card itself.