Bcache causes data lose.

Bcache causes data lose.

[Yuxuan Shui]

Hi,

I'm trying bcache for the past two days. And I observed several data
corruptions.

This happens with both writeback and writethrough cache. After reboot,
I saw my filesystem corrupted. With writeback cache, there's no way to
recover it. With writethrough cache, after I detach the backing device
from the cache set, I can mount my filesystem again. So I think this
is cleanly a bcache bug.

I don't know if above information is enough, if you need to anything,
I'll do my best to provide

Regards,
Yuxuan Shui.

Re: Bcache causes data lose.

[Zachary Palmer]

Which kernel are you running?  I had very similar problems with all 3.11 
kernels below 3.11.5; kernel 3.11.5 works just fine for me.  The 3.10 
kernels also seemed to be okay, but the mailing list has suggested that 
they may have some "time bomb" bugs, so you should probably be running 
3.11.5 or newer.

Cheers,

Zach
> Hi,
>
> I'm trying bcache for the past two days. And I observed several data
> corruptions.
>
> This happens with both writeback and writethrough cache. After reboot,
> I saw my filesystem corrupted. With writeback cache, there's no way to
> recover it. With writethrough cache, after I detach the backing device
> from the cache set, I can mount my filesystem again. So I think this
> is cleanly a bcache bug.
>
> I don't know if above information is enough, if you need to anything,
> I'll do my best to provide
>
> Regards,
> Yuxuan Shui.
> --
> To unsubscribe from this list: send the line "unsubscribe linux-bcache" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

Re: Bcache causes data lose.

[Yuxuan Shui]

I was running 3.11.6 when I first saw this. And I tried 3.12, same
problem. Only with 3.12 when this happen there's a line like "bcache
spotted bad key" in dmesg.

On Thu, Nov 7, 2013 at 1:06 PM, Zachary Palmer <zep_bcache-J5qI5MFTcs8@public.gmane.org> wrote:
> Which kernel are you running?  I had very similar problems with all 3.11
> kernels below 3.11.5; kernel 3.11.5 works just fine for me.  The 3.10
> kernels also seemed to be okay, but the mailing list has suggested that they
> may have some "time bomb" bugs, so you should probably be running 3.11.5 or
> newer.
>
> Cheers,
>
> Zach
>>
>> Hi,
>>
>> I'm trying bcache for the past two days. And I observed several data
>> corruptions.
>>
>> This happens with both writeback and writethrough cache. After reboot,
>> I saw my filesystem corrupted. With writeback cache, there's no way to
>> recover it. With writethrough cache, after I detach the backing device
>> from the cache set, I can mount my filesystem again. So I think this
>> is cleanly a bcache bug.
>>
>> I don't know if above information is enough, if you need to anything,
>> I'll do my best to provide
>>
>> Regards,
>> Yuxuan Shui.
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-bcache" in
>> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>
>

Re: Bcache causes data lose.

[Kent Overstreet]

On Thu, Nov 07, 2013 at 10:39:39AM +0000, Yuxuan Shui wrote:
> Hi,
> 
> I'm trying bcache for the past two days. And I observed several data
> corruptions.
> 
> This happens with both writeback and writethrough cache. After reboot,
> I saw my filesystem corrupted. With writeback cache, there's no way to
> recover it. With writethrough cache, after I detach the backing device
> from the cache set, I can mount my filesystem again. So I think this
> is cleanly a bcache bug.
> 
> I don't know if above information is enough, if you need to anything,
> I'll do my best to provide

That's not good :/ Can you tell me everything that's in your setup? md/lvm/what
drivers you're using for your drives/anything else that might be relevant...

Re: Bcache causes data lose.

[Yuxuan Shui]

On Thu, Nov 7, 2013 at 8:22 PM, Kent Overstreet <kmo-PEzghdH756F8UrSeD/g0lQ@public.gmane.org> wrote:
> On Thu, Nov 07, 2013 at 10:39:39AM +0000, Yuxuan Shui wrote:
>> Hi,
>>
>> I'm trying bcache for the past two days. And I observed several data
>> corruptions.
>>
>> This happens with both writeback and writethrough cache. After reboot,
>> I saw my filesystem corrupted. With writeback cache, there's no way to
>> recover it. With writethrough cache, after I detach the backing device
>> from the cache set, I can mount my filesystem again. So I think this
>> is cleanly a bcache bug.
>>
>> I don't know if above information is enough, if you need to anything,
>> I'll do my best to provide
>
> That's not good :/ Can you tell me everything that's in your setup? md/lvm/what
> drivers you're using for your drives/anything else that might be relevant...

I'm running a almost fresh Archlinux installation, first with a 3.11.6
kernel then updated to 3.12.

My SSD drive is Samsung 840EVO 120G. I'm not using md or lvm, just
btrfs on top of bcache.

Re: Bcache causes data lose / BTRFS

[Rolf Fokkens]

On 11/08/2013 05:13 AM, Yuxuan Shui wrote:
> My SSD drive is Samsung 840EVO 120G. I'm not using md or lvm, just
> btrfs on top of bcache.
While testing bcache in Fedora I noticed BTRFS reporting checksum errors 
in kernels <= 3.11.4. Not knowing if it was a BRTFS issue or a bcache 
issue and the fact that there's no proper fsck for BTRFS I stopped using 
BTRFS.

Could BTRFS be a good way to test bcache because it has checksumming? Or 
will it only complicate testing?

Rolf