How to Repair Corrupt master file table (windows 2000 domain controller will not boot)

One of my clients called me reporting that their server stopped responding and so they rebooted. Upon reboot, the server seemed to hang forever and would not actually boot. I attempted to boot into safe mode, nothing, active directory restore mode, nothing, recover console, nothing.

Next step is to boot the windows 2000 server CD-ROM and run a repair (note be very careful about trying this on a domain controller, PLEASE BACKUP YOUR SYSTEM BEFORE TRYING ANY of THIS STUFF, USE GHOST AND BACKUP THE ENTIRE DRIVE TO AN IMAGE FILE SO THAT YOU HAVE A BASELINE TO GO BACK TO)

Even running the command from the recover console would not allow me to boot and do anything. Never before had I been so completely locked out of a system.

Now it was time to bust out the big tools.

Located at www.bootdisk.com (a really great web site for computer techs) there is a boot disk called NTFS for DOS. This is a great freeware utility that will allow you to create a floppy disk, insert into the windows computer with NTFS on it and then view the files, run a checkdisk, and defragment the drive. All this from a dos console on a NTFS formatted drive. I put the disk in and launch the second option (checkdisk utility) this is the one that wouldn't work via booting to the windows CD-ROM. http://bootdisk.com/ntfs.htm

While running the check disk utility, in short order it returns the error message, "corrupt master file table" check disk aborted. Basically, its a corrupt MFT. Usually when you see this error message your done. That would be a reformat and start over type of error. In this particular case, that would be a huge bummer as this is our primary domain controller, exchange server, well actually our only domain controller and email server. No users can log on to their computers, no mail is flowing into the building, no one can access any files on the file server. Basically, the entire company is at a standstill until this issue is resolved.

The next step I took was to pop another separate hard drive into the server. I then installed a new copy of windows 2000 server. This could have been 2000 professional, or windows XP. Basically we just want an Operating system that has the ability to read NTFS file system. The idea being that we boot to the new OS, and then run the windows checkdisk on the old partition and try to repair the corrupted MFT file. If all goes well there, we should then be able to boot into the server and life should be good.

30 minutes later, the new windows 2000 server was installed, now for the big play. does the checkdisk resolve the problem or are we totally screwed?!?!?!?

OHHH problem, it seems that keeping the old raid controller in the system was a mistake, my server installation failed because of the old disk was still in the server. Shut off the power, yank the 3ware 9000 series raid controller out of the box and start over with a new hard drive just to be sure that the hard drive was not bad in some way. ok now lets wait another 30 minutes to see if this install will take this time.

I got the system up and now its time to run checkdisk to see if this will fix the problem that we encountered earlier in the process


from the command prompt run
chkdsk /f e: (this should be the drive letter of your old partition)

BUMMER!!! this fixed some problems with the file system so I tried to boot into the old partition, No luck. Next we need to check the registry for problems that may be stopping us from booting.

use the attached chkreg utility found at chkreg.exe

the syntax is

chkreg /f e:\winnt\system32\config\system /r
(returned no errors)
chkreg /f e:\winnt\system32\config\software /r
(Invalid Base Block CheckSum...Fixed, Sequence Numbers do not match ... Fixed)
chkreg /f e:\winnt\system32\config\SAM /r
(no errors)
chkreg /f e:\winnt\system32\config\SECURITY /r
(no errors)

SUCCESS!!! HURRAY!!! WE ARE THERE!!!! YESSSSSSSSS!

ok now its a happy day. Hope you have a happy day too.