Understand rootFS A/B

Autonomous Machines Jetson & Embedded Systems Jetson Orin NX
1

Hi,
I wish to use the rootFS redundancy feature but there are few points left unclear to me:

  1. I read the following sentence in the dev guide:

As a failover mechanism, if the current active rootfs were damaged or corrupted, and then can’t boot up the system, after automatically rebooting 3 times (customizable), the device will failover to the other available unused rootfs to boot up. If both rootfs A and B are unbootable, the device will boot into the recovery kernel image.

My question is why I need an additional rootFS if I have a recovery kernel image available?
Maybe I don’t understand the meaning of “recovery kernel image”:

  • Does it provides a “rescue” shell prompt enabling me modifying the corrupted rootfs?
  • Does it contains the needed “basic” binaries for me to modify the corrupted rootfs, or maybe I can use the ones existing on that corrupted rootfs (its /bin , /sbin directories)?

  1. Can I use different “types” of rootfs for my A/B rootfs? for example rootfs A will be the sample one and rootfs B will be a minimal one without a GUI?

  2. Can I access (root priviliges) the rootfs files that is not being loaded? (the system booted with rootfsA and I use root account to access rootfsB internal files)

  3. I read this line:

The Bootloader Update Payload (BUP) is the payload that is applied by the update engine during an update

and my question is why I need some special bootloader update mechanism if I can just update my bootloader image on my host and flash it directly to its partition using the flash.sh script or the initrd method.

Thanks

3

hello BSP_User,

please refer to developer guide, Root File System Redundancy.
>>Q1
it’ll boot into bash, only few commands are available.

>>Q2
yes. you may customized that by your own use-case.

>>Q3
no.

>>Q4
A Bootloader Update Payload (BUP) is used to implement Bootloader update.

4

Thank you for your answer. I just wish to clarify few points:

Q1:

  • If I have a rescue bash shell, what is the use case for a second “backup” RootFS?
    can I just use this shell to repair my rootFS?

  • In addition, I flashed my target, executed rm -rf /* and got kernel panic again and again without rescue bash shell. Are you sure that in case of rootFS corruption I get a bash rescue shell?

Q3:
I installed the A/B mechanism and successfully edited both slots rootFS files using sudo. While using the first rootfs, the second one is just mounted as regular disk, hence available for modifications.

5

please see-also Topic 197124.

6

Thank you

8

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.