Migration from the KQ Implementation

Gordan Bobic gordan.bobic at gmail.com
Wed May 4 21:05:20 EDT 2011


On 05/05/2011 02:00, Jason J. W. Williams wrote:
>> Considering I was talking to somebody at the place where I work last week
>> about a case where ZFS wasn't recoverable, I can only conclude that you've
>> been lucky. The only way the file system got salvaged in the case I'm
>> talking about was by enlisting the services of one of the original ZFS
>> developers and some of his custom tools for recovering data off a ZFS file
>> system. It does happen, and tweaking zdb isn't always sufficient.
> I'd say they were unlucky. Only cases I've encountered that ended in
> that kind of recovery involved bad storage controllers, and fsck
> wouldn't have helped (and I'm aware of most of those cases). A custom
> recovery job would have been required anyway and that's the point, the
> things you want fsck for ZFS does already. Fsck is a very limited
> tool.

I would argue that fsck is implicitly less limited than whatever the 
kernel module can do on the fly. fsck wouldn't have to make as many 
assumptions when it comes to trying to pick up a broken pool and running 
with it.

Gordan



More information about the zfs-discuss mailing list