Corrupt Files Issue

[CJ Mar]

I'm dealing with a customer with a strange issue that is intermittant and
doesn't happen often so it is hard to troubleshoot. Looking for some advice
or suggestions. Here is a rundown of the server first.

They have a Compaq Proliant ML350 G2 system with Windows Server 2003
installed. They have 4 drives in the server using a built in compaq scsi
array. There are 3 drives in one array containing 2 partitions (C and D)
and another drive in an array by itself which is (E).

Here is the problem. In the space of 6 months they had the same problem
happen twice. The server began reporting that random files were corrupted
and were inaccessable which came from the D partition. These files are all
different and random files that users in the network have put in different
directories on the server that are shared to everyone. A balloon would pop
up in the task tray and state the files were corrupted and when you try to
access them they won't work. We tried a restore from a backup and they
would not replace. We had to setup a scan disk on the D drive and when the
server rebooted and did the scan disk it deleted the files. Once the server
was up and runnning again we were able to restore the missing files
sucessfully.

What would cause these files to become corrupt? Is this a hardware issue or
a Windows Server 2003 issue? We setup a defrag for a weekly scan incase the
files were becomming fragmented and causing the issue. We also tried scan
disk and other utilities on the drive which show that the hardware seems to
be fine.

Help! Any suggestions?

[Jon Finley]

There was an issue (just lately) that addressed a problem with NTFS and compression that could cause data corruption.

A patch that was released a while back was being re-worked and re-released.

Make sure that your Server is up-to-date on updates/patches.

http://www.techweb.com/wire/security/193002107

http://www.microsoft.com/technet/security/bulletin/ms06-049.mspx?pubDate=2006-08-08


Jon

[Joel Belizario]

I'd recommend doing an extended test of the HDDs using Insight Diags (offline preferably using the SmartStart CD) as I have run into a similar situation before with a non-HP server.

It was caused by a faulty HDD in a RAID 5 array (3 disks) which did not report as faulty in diags, only found out after recreating the array and waiting for the sync to complete - only then did one of the drives show up as failed.

[Gene Laoyan]

If your the disk/volume is usung NTFS compression, undo it.

I recommend running Spinrite 6.0 on the disks in question. Hard drives do tend to grow errors.