I’m testing deployments using Mender in a Jetson TX2 device (using Jetpack 4.6.1, ota tools version R32.6.1) following the tutorial available in Nvidia Jetson L4T Integration - Debian family - Mender Hub. I have already created a working artifact and I was able to deploy it correctly on another device.
Now, I’m testing scenarios where errors are introduced in a controllable way. In one of them, the non-bootable artifact has been created by removing Image and zImage files from the boot folder in the OS snapshot. According to the Nvidia documentation, if the device fails to boot to the updated slot, it switches back to the previous slot, however in this test the device did not boot. Should I introduce any step in the procedure described here to permit a rollback in this kind of situation?