We are currently exploring computer vision-based development for Jetson TX1 board. Aside from a lot of technical questions, I would like to know about the procedure for deploying our developments in multiple TX1 boards.
I assume one could develop locally within a host machine and then test-debug over a Jetson TX1 Development Kit. Once an application is ready for a production environment, it is expected to install it in several Jetson modules with a suitable carrier board.
Can anyone confirm whether is this the correct way? If so, each board must be prepared individually or it can be flashed by means of disk images?
Any comments about your own experiences will be more than welcome.
You can clone partitions on a Jetson. If the partition is the root partition then you can loopback mount this, edit this, and then use the flash app to flash using this image rather than creating a new image. See:
This URL is for a TK1, but most of the concept applies to a TX1:
Note that an image intended for a particular release (such as R24.2) may require the Jetson to have been flashed once with that release before the root partition is compatible. I mention this because running the flash.sh script takes care of this when naming the “-r” re-use option, but if you directly run some of the tools to flash only a single partition and the other parts are from different releases you might have issues.
The cloned partition is loopback mountable as ext4. This can be edited before flashing with re-use and the edits will carry forward into the restored clone.
Thank you for your answer, I will try it soon.