BTRFS warnings, next steps?

[ruvil]

Hi

 

So i noticed in my system log that i got the following error message for one of my disks:

Quote

[771094.707666] BTRFS warning (device md3): csum failed root 5 ino 193741 off 4879454208 csum 0x35b88e06 expected csum 0x80435938 mirror 1
[771094.991283] BTRFS warning (device md3): csum failed root 5 ino 193741 off 4879454208 csum 0x35b88e06 expected csum 0x80435938 mirror 1
[771095.386307] BTRFS warning (device md3): csum failed root 5 ino 193741 off 4879454208 csum 0x35b88e06 expected csum 0x80435938 mirror 1
[771095.386449] BTRFS warning (device md3): csum failed root 5 ino 193741 off 4879454208 csum 0x35b88e06 expected csum 0x80435938 mirror 1
[775862.193890] BTRFS warning (device md3): csum failed root 5 ino 193741 off 4879454208 csum 0x35b88e06 expected csum 0x80435938 mirror 1
[775862.569906] BTRFS warning (device md3): csum failed root 5 ino 193741 off 4879454208 csum 0x35b88e06 expected csum 0x80435938 mirror 1
[775862.570223] BTRFS warning (device md3): csum failed root 5 ino 193741 off 4879454208 csum 0x35b88e06 expected csum 0x80435938 mirror 1
 

 

What should the next steps be for me to not complicate things further?

 

SMART data looks good, no issues there from what i can see.

[JorgeB]

There's some data corruption, it can be caused for example by bad RAM (if not using ECC) or other issues, first step would be to run memtest (again if not using ECC), second is to run a scrub, then look on the log to find out which files are corrupt.

[ruvil]

Interesting thank you!

 

I did run a scrub, fortunately only one file was affected but i also got the following error message:

Quote

[808975.393115] BTRFS error (device md3): bdev /dev/md3 errs: wr 0, rd 0, flush 0, corrupt 4, gen 0
[808975.393116] BTRFS error (device md3): unable to fixup (regular) error at logical 3068779085824 on dev /dev/md3

 

This is the log from the BTRFS scrub:

Quote

scrub status for 9a9fde6a-cc62-4b19-b099-787d828c3a5a
	scrub started at Fri Oct 12 00:08:39 2018 and finished after 05:16:35
	total bytes scrubbed: 1.82TiB with 5 errors
	error details: csum=5
	corrected errors: 0, uncorrectable errors: 5, unverified errors: 0

 

 

[JorgeB]

7 minutes ago, ruvil said:

BTRFS error (device md3): unable to fixup (regular) error at logical 3068779085824 on dev /dev/md3

That's normal, it just means btrfs can't fix the error since there's no redundancy.

 

You just need to figure out why the corruption happened in the first place, if that disk was disabled at some point in the past during a write it's a possible explanation, if not there's likely some hardware problem on the server.