BTRFS corruptions counter

BTRFS corruptions counter

[Todor Ivanov]

Hi,

We decided to give BTRFS a try. We find it very flexible and generally
fast. However last week we had a problem with a Marvell controller in
AHCI and one BTRFS formatted hard drive. We isolated the problem by
relocating the disk to an Intel contoller (SATA controller: Marvell
Technology Group Ltd. 88SE9172 SATA 6Gb/s Controller (rev 11) had a
lot of problems and I managed to overcome them by passing
libata.force=noncq for a couple of weeks, until it failed generally).
Now that the disk is on the intel contoller it works fine and data is
intact, but whenever I do mount or unmount I see:

Aug 10 18:34:47 MyServerHere kernel: btrfs: bdev /dev/sde1 errs: wr 0,
rd 0, flush 0, corrupt 166, gen 0

The problem is "corrupt 166"
Sofar I did:

1. btrfsck and it reported everything is OK
2. btrfs scrub start /dev/sde1 and it again reported all is OK

Am I missing something in my filesystem checks or there is another way
to zero the corrupt counter?

Regards,
Todor

Re: BTRFS corruptions counter

[Stefan Behrens]

On 08/12/2013 18:57, Todor Ivanov wrote:
> Hi,
>
> We decided to give BTRFS a try. We find it very flexible and generally
> fast. However last week we had a problem with a Marvell controller in
> AHCI and one BTRFS formatted hard drive. We isolated the problem by
> relocating the disk to an Intel contoller (SATA controller: Marvell
> Technology Group Ltd. 88SE9172 SATA 6Gb/s Controller (rev 11) had a
> lot of problems and I managed to overcome them by passing
> libata.force=noncq for a couple of weeks, until it failed generally).
> Now that the disk is on the intel contoller it works fine and data is
> intact, but whenever I do mount or unmount I see:
>
> Aug 10 18:34:47 MyServerHere kernel: btrfs: bdev /dev/sde1 errs: wr 0,
> rd 0, flush 0, corrupt 166, gen 0
>
> The problem is "corrupt 166"
> Sofar I did:
>
> 1. btrfsck and it reported everything is OK
> 2. btrfs scrub start /dev/sde1 and it again reported all is OK
>
> Am I missing something in my filesystem checks or there is another way
> to zero the corrupt counter?

These counters are not reset automatically. To do this manually, run 
this command:

     btrfs device stats [-z] <path>|<device>
         Show current device IO stats. -z to reset stats afterwards.

If you don't have this command, you can find instructions how to compile 
updated btrgs-progs tools here:

https://btrfs.wiki.kernel.org/index.php/Btrfs_source_repositories#btrfs-progs_git_repository

Additionally, read the section "build dependencies" which contains a 
list of packages that are required to build the tools.