drive failure, now 'cannot alloc 494581644 bytes for inoinfo' and 'bad inode number 3556352 to nextinode'

[Will Saxon]

Hello,

Today, I had a drive fail in a large raid 5 array hosted off of a ciss device. I removed the bad drive but before I was able to get another drive swapped in, another drive seemed to just lose power and then the system panicked.  

I swapped in a spare drive and restarted the system. The controller said the array was 'ok' and would recover. However, FreeBSD is having a hard time fsck'ing the partition.

At the time of the crash, I was using FreeBSD 5.2.1-RC2, however I have since updated to 5.2.1-RELEASE-p4. The array has 1 partition and 2 slices - one 4GB swap slice and 1 400+GB storage slice. This was being used in a samba pilot and had the softupdates, suiddir, noexec, nodev and acl options enabled.

Basically, in phase 1 fsck -y complains that it cannot allocate enough bytes for inoinfo, then there are a bunch of the following:

UNKNOWN FILE TYPE I=#######
UNEXPECTED SOFT UPDATE INCONSISTENCY

then finally:

fsck_4.2bsd: bad inode number 3556352 to nextinode

Any ideas?

-Will

_____________________________________________
Will Saxon
Systems Programmer - Network Services
Department of Housing and Residence Education
University of Florida
Email: wills at housing.ufl.edu
Phone: (352) 392-2171 x10148