Windows\NTDS\edb.log corruption.

I have 3 DCs and 2 of them experience Windows\NTDS\edb.log corruption, this week and I am afraid that my last DC will be next. The first DC is not booting in normal mode so I ended up forcing a demotion and then bringing it back intp normal mode, so I ended up forcing a demotion and then bringing it back into the domain. It looks to come back fine and everything ran ok for the past week. But today, my 2nd DC started to have same errors, that was, Directory Services events 465, etc. The first DC that failed is Windows 2003 64bit and todays DC is Windows 2003 32 bit and it runs as a virtual machine under the first failing DC. The one DC that has not failed is Windows 2003 32bit on a real machine with no virtual's under it. Can anyone tell me what should I do now? Thanks.

You can try to run dcdiag and netdiaf from support tools. After that check the replication errors, disk space or errors, memory, etc. If everything is working fine, then you will need to run an offline defrag of AD Database.

Did you try to run a Semantic Database analysis yet? Check the article made on, how to complete a semantic database analysis for the Active Directory database by using Ntdsutil.exe from here - http://support.microsoft.com/kb/315136

You need to check the root cause of this issue. Are you using professional hardware with ECC memory? Is you disk controller cache use ECC memory? Have you checked the disk controller has a battery backup, or is write cachning disabled? Incase you are not using ECC memory, then try to run memtest86 on the machine to find out if your memory is faulty. The disk controller could also be at fault, so check it as well.

I tried everything out there to fix the AD errors, but nothing had worked. Bad thing is, tape backups had failed & was just installing new tape drive when AD became corrupted.

I ran the following commands:
esentutl /g “c:\winnt\ntds\ntds.dit” /!10240 /8 /v /x /o
might fail, but do next step anyway
esentutl /p “c:\winnt\ntds\ntds.dit” /!10240 /8 /v /x /o
Deleted log files
Rebooted server normally

This fixed the problem.