Bcache (device?) failure

Bcache (device?) failure

[David H. Rhodes Clymer]

Speaking of bcache integrity...This morning, bcache started indicating
that it was encountering errors writing to the drive. 

For some stupid reason, I rebooted, and on boot, it cannot find the
volume group (I was running LVM on bcache) that my root fs is on,
finally dropping to the initramfs shell.

I then find that attempting to register my bcache component devices
fails:

(initramfs) echo /dev/sdb1 > /sys/fs/bcache/register
sh: write error: invalid argument
(initramfs) ls /sys/fs/bcache/
register         register_quiet

This happens for both the cache device and the backing store. I realize
that collecting information prior to a reboot would have been much more
useful...sorry. 

Is there any simple reason why I would be seeing this behavior? I assume
that either both devices are corrupt, and no longer seen as bcache
formatted devices. Does that seem likely?

-davidc

Re: Bcache (device?) failure

[Kent Overstreet]

On Fri, Jan 25, 2013 at 11:30:22AM -0500, David H. Rhodes Clymer wrote:
> Speaking of bcache integrity...This morning, bcache started indicating
> that it was encountering errors writing to the drive. 
> 
> For some stupid reason, I rebooted, and on boot, it cannot find the
> volume group (I was running LVM on bcache) that my root fs is on,
> finally dropping to the initramfs shell.
> 
> I then find that attempting to register my bcache component devices
> fails:
> 
> (initramfs) echo /dev/sdb1 > /sys/fs/bcache/register
> sh: write error: invalid argument
> (initramfs) ls /sys/fs/bcache/
> register         register_quiet
> 
> This happens for both the cache device and the backing store. I realize
> that collecting information prior to a reboot would have been much more
> useful...sorry. 
> 
> Is there any simple reason why I would be seeing this behavior? I assume
> that either both devices are corrupt, and no longer seen as bcache
> formatted devices. Does that seem likely?

Yeah... you'd have to check the dmesg log to be sure, but the most
likely cause for that is bcache isn't finding a bcache superblock :/