fsck crash: bad inode number to nextinode

[Greg Eden]

On 27 Nov 2006, at 20:22, Eric Anderson wrote:

> On 11/24/06 03:31, Greg Eden wrote:
>> Hello,
>> I'm try to recover a RAID5 volume which was badly corrupted when  
>> a  drive was removed during a rebuild. It contained about 1 TB of  
>> data  and was formatted with default values under FreeBSD 6.0-R.
>> I have used dd to image the drive onto another volume and am  
>> mounting  it with mdconfig so I can work on that an not cause  
>> futher damage.  However when I run fsck_ufs on the /dev/md0  
>> partition it eventually  crashes out during Phase 1 with
>> UNKNOWN FILE TYPE I=42151497
>> UNEXPECTED SOFT UPDATE INCONSISTENCY
>> CLEAR? yes
>> fsck_ufs: bad inode number 42158080 to nextinode
>> Is it possible to work around this to get fsck to complete?
>> It is possible to mount the partition and some of the data is  
>> there,  however most of it is not.
>> Thanks in advance for any help. I have previously posted to  
>> freebsd- questions without a response.
>
>
> I've seen this before with really badly UFS filesystems, where the  
> cylinder groups were mangled.  I couldn't think of a good way to  
> have fsck fix this, since you can't really guess at the inode  
> information, and so the only option is really to just 'delete' the  
> inode information, but that wasn't clear to me how to do that safely.

OK. I had a feeling fsck wasn't going to save me this time :(

> You would probably be best served by running one of the various  
> tools (in source and also in ports) that try to recover files  
> themselves from a dd'ed image.

Do you have any specific recommendations? a search of freshports.org  
revealed 'magicrescue' and 'foremost' as likely looking rescue  
utilities. Is there anything else?

Thanks for the pointers!

best.
greg.