Fsck error at startup on Apalis TK1

Technical Support
, ,
1

Once in a while, when booting the device, I see on the serial console:

run-parts: /etc/update-motd.98-fsck-at-reboot exited with return code 1

Which I assume means there is a corruption on the mmc storage.

A few questions:

  1. How do I run force fsck at startup?
    The standard methods of touch /forcefsck or shutdown -rF or tune2fs -c 1 /dev/mmcblk1 don’t seem to work. Right now, fsck seems to run at startup only every N reboots, which makes it hard to detect & reproduce the problem.
  2. How do I fix errors manually? I don’t see a way to run fsck in the bootloader (which I could run manually)
  3. How do I automatically fix errors when the device encounters such a situation in the field (where manual intervention is not possible)

Thanks in advance.

2

What exact hardware and software versions of things are you talking about? Is this with our regular BSP demo image, the mainline image or running JetPack? Could you please also provide full log file(s) when the issue occurs?

3

This is with the Jetpack installed as per https://developer.toradex.com/knowledge-base/installing-nvidia-jetpack-with-l4t-on-apalis-tk1.
Which logs can I provide? dmesg?

4

yeah, provide the dmesg log.

5

link text

6

Attached dmesg

7

Hi Sagi

It looks like you are running a custom kernel and/or device tree. Can you reproduce the same issue with our stock BSP artefacts? Are you using some kind of NFS configuration (idmapd is typically used for such use cases)?

BTW: You should not be running 2.7.4 being considered an obsolete beta version by now but rather upgrade to 2.7.5 at your earliest convenience.

Cheers

Marcel

8

We are using Tegra L4T 3.1 built as per toradex instructions.
No use for the basic BSP.
Accordingly we would still need to understand how to troubleshoot and fix this on apalis Tk1 with L4T.

9

As NVIDIA’s L4T is based on Canonical’s Ubuntu ARM I guess the same principles apply.