Master File Table ($MFT) Corrupt

[Guest]

We run an application that writes a large number of small (~20KB) files to
large volumes (~1 TB). We have started to see the $MFT corrupted on a regular
basis on many machines with various disk controllers (SCSI/SATA/IDE.)

In each of these situations we run chkdsk /x on the drive in question and
the problem is resolved. However, after running our application for another
week or so we are faced with the same situation.

Presumably this is caused by the way in which we are writing to the drive
(not the hardware,) since we have seen this problem across multiple hardware
platforms.

Upon inspection our $MFT is slightly fragmented (usually 2-3 fragments) and
is within its reserved space on the disk.

Has anyone seen similar problems?

The chkdsk output looks almost identical in each situation:
Checking file system on S:
The type of the file system is NTFS.
Deleted corrupt attribute list entry
with type code 176 in file 0.
Unable to locate attribute of type 0xb0, lowest vcn 0x0,
instance tag 0x0 in file 0xf.
The attributes with instance tags 0x11 and 0x0 have different
type codes 0x80 and 0xb0 respectively in file 0x0.
The attribute of type 0x80 and instance tag 0x11 in file 0x0
has allocated length of 0x510d88000 instead of 0x289000.
Deleted corrupt attribute list entry
with type code 128 in file 0.
Unable to locate attribute with instance tag 0x11 and segment
reference 0x1000000000000. The expected attribute type is 0x80.
Deleting corrupt attribute record (128, "")
from file record segment 0.
Unable to locate attribute with instance tag 0x0 and segment
reference 0xf00000000000f. The expected attribute type is 0xb0.
Deleting corrupt attribute record (176, "")
from file record segment 15.
Deleting orphan file record segment 15.
Cleaning up minor inconsistencies on the drive.
CHKDSK is recovering lost files.
Inserting data attribute into file 0.
Inserting data attribute into file 15.
The MFT mirror is different from the MFT.
Correcting errors in the Master File Table (MFT) mirror.
Correcting errors in the master file table's (MFT) DATA attribute.
Correcting errors in the master file table's (MFT) BITMAP attribute.
CHKDSK discovered free space marked as allocated in the volume bitmap.
Windows has made corrections to the file system.


Thanks,
-Jason

 

[kalosoft]

I have the same problem - about 200 GB of small files (avg. 20kb) on
NTFS volume under Windows XP, which failed today.
I tried several data recovery solutions with no success Some of them
wera able to find few files, some more, but partial recovery is not an
option in my case, so I'll have to reformat it and restore from
backups.
It's the second time this happens, so maybe the right question is - how
to store such a large number of files without a risk? Does any
alternative high-performance redundant filesystems for Windows exist
which could handle this situation?