BTRFS warning (device mmcblk0p1): csum failed

My Turris Omnia take much longer to start up than usual. Today I found a lot of erros like this:

2019-03-15 19:41:38 warning kernel[]: [ 1815.289749] BTRFS warning (device mmcblk0p1): csum failed ino 262775 off 1871872 csum 2186375717 expected csum 3312312294
2019-03-15 19:41:38 info kernel[]: [ 1815.289867] Core dump to |/usr/lib/systemd/systemd-coredump 31785 0 0 7 1552675298 foris pipe failed

I tried this:

[root@turris lxc] # btrfs scrub start -B /
ERROR: scrubbing / failed for device id 1: ret=-1, errno=5 (I/O error)
scrub canceled for 4f32477e-ac11-415f-9699-60e865844b2f
        scrub started at Fri Mar 15 20:03:03 2019 and was aborted after 00:00:07
        total bytes scrubbed: 61.52MiB with 480 errors
        error details: super=2 verify=1 csum=477
        corrected errors: 0, uncorrectable errors: 478, unverified errors: 0

This is really annoying. I bought this device in November 2016 because I thought that it is a device of high quality and good IPv6 compability.

Now I will replace it with some cheap router.

Cheap router won’t tell you something is wrong:D

Try this: Netdata error with BTRFS partition mounted in /srv
I also had problems after some power failures.

I have replaced the router by a Fritz!Box 7530. And for me this is the end of the story.

Sorry but am I reading your post correctly that you’ve used lxc on internal flash??

Yes - but the the external flash was not affected but the internal. So in the end I was not even able to boot the router.

I don’t understand the “yes - but” combination. To be clear, systems in LXCs tend to do lots of writes, wearing down (parts of) the flash after some time. That’s why it’s strongly recommended not to run them in Omnia’s internal flash, over long term at least.

What I meant is: yes - I used LXC containers but on an external SSD card. There was an instruction how to remove one Wifi card and put it on another place and put the SSD card inside. The default config was in a way you could add some LTE SIM card as fallback. I did bought a mSATA SSD and deployed LXC containers there.

But this card was not affected with the problem. The problem was the internal flash (where I did not deploy any containers).

Right, thanks for clarifying that… so LXC don’t seem to be the cause.

I wouldn’t be so sure. We don’t know, how he has it mounted and if he wasn’t using Storage plugin, which is in Foris, it could happen that it couldn’t find mSATA SSD and then it was writing to eMMC.

Yes, of course… I meant that the little information shared so far doesn’t seem to point to LXC (to me) or anywhere actually.