What is the best practices for performing an A/B rootfs update? Following this documentation, I am able to manually generate a rootfs. My ideal path would be to take this and then deploy it to the inactive partition and then call nvbootctrl to switch over the active partition. From my understanding, this should work, correct?
I’ve read about capsules and BUPs, however I’d prefer to stay away from that solution from now because we’d like to adapt our in-house deployment system to update the rootfs for the jetsons only for OTA updates, and do not foresee the need to update jetpack versions, kernels or DTBs.
However, when I do follow this path, after I extract the sample_fs.tar.gz generated by nv_build_samplefs.sh
, i just get a blank screen for next boot and never actually see the other partition boot. FWIW, I have validated that my A/B partitioning scheme is setup correctly, as i can swap between the partitions using nvbootcrl (as well as verify that they have been set using the same tool) before I modify either of the rootfs’s.
Is using nv_build_samplefs.sh
the preferred method, or should I use the flash tool to take a recovery image of one partition and then use that to flash the other?