Looks like we have experienced our first catastrophic failure that we were not able to bring back from an SP image.

Every time we imaged the new drives that went into the server, we received CHKDSKs during the inital bootup phases with an inability to fix the errors due to insufficient disk space. In some cases the OS was able to boot, but we were greeted with a plethora of NTFS Event ID 55s.

The images themselves verify fine in SP GUI.

Any thoughts?

My suspicion is that the MFT was gradually degrading until we shutdown the server to install some new RAM.

The server was running an SRCS16 Intel RAID controller with two arrays defined on it. Peformance when restoring the OS partition varied from .27MB/sec to 4.75MB/sec. Whether we were in WinPE or Legacy, loaded the RAID controller drivers or not, there did not seem to be a correlation between the performance we were getting on the restore. All images were 128bit AES password protected.

So, in this case, we will be using a Swing Migration to bring the SBS back online.

*sigh*

There are a couple of blog posts on this particular struggle too.

Philip

MPECS Inc.

http://blog.mpecsinc.ca

I've never seen NTFS corruption except with faulty hardware. Can you exclude bad hardware as culprit? Looks like at one point something bad happened to your server while Windows kept running but the image taken is not bootable anymore because the original server isn't either. This would be a case for 2nd stage tape backups taken monthly with 12 months full of backups to recover from.

I'd say check and double check hardware then rebuild whole server and restore/reinstate all data that you can salvage from SP backups. 

This particular box originally had a faulty motherboard. It was replaced by another faulty motherboard which in turn took out some of the RAM sticks.

This operation was to replace those faulty sticks.

So, yes, there was definitely a problem at the hardware level. Those problems unfortunately got passed along in the images.

Since I have the bad box here, once we have finished swinging their SBS domain to new hardware, I will be reloading an OS on the box and stress testing it.

Tough lesson. Even if we used the Swing method to put a fresh install of SBS on the problematic box, there would be no guarantees that it would remain solid in production.

Thanks for the reply.

Philip

You could also try mounting the Backup Image as a writable volume and then running chkdsk /f on a good machine.  Once the chkdsk has completed to you can save the changes made by chkdsk to an incremental file and then use it to restore. 

STC