Software

Ask something
Discussion Locked

Corrupt Exchange Information Store

0 helpful
5 replies

by evogel

Closed to new repliesThis discussion is closed. It stays readable, but it takes no new replies.
I have a client who has a Windows SBS 4.5 server with SP6a installed and Exchange 5.5 SP3. Circular logging is not enabled on the exchange server. The priv.edb is about 4GB and it resides on a 30GB hard drive with plenty of free space. The serveralso is using Backup Exec 8.0 for SBS for its backup software (upgraded about three months ago) and Trend Micro Office Scan as its antivirus software (changed from Norton Antivirus at the time of the server purchase). The server is a Dell Power Edge 2400 that was purchased at the beginning of this year and there are no other servers on the network. At random intervals (ranging from 3-6 months) the information store on this server stops with error (when you try to start it), "server specific error 4294966746" and entries in the event viewer indicate that there is a corruption in the exchange information store. If I try to resolve this error via following the steps in KB article Q219419 then the server will lock up when I get to the hard recovery step of the priv.edb (none of the steps before that resolve the problem) and to make a long story short, I will just have to restore from backup (a full backup of exchange is run every night). This problem happened on the old server as wellas the new and it happened when Exchange still had SP2 on it. When data was transferred from the old server to the new the exmerge utility was used to bring all of the users data into a new exchange database. There is also a script that runs each night that stops the antivirus software prior to running the backup (it happened before this script was implemented). There are no entries in Dr.Watson or the event viewer that give a clue as to why this is happening. I would like any suggestions that would help me find out why this is happening or how this can be prevented from happening in the future. Thanks.

5 replies

by mebudman

The dreaded 4294966746 database inconsistent, run eseutil /p, as only resort. If this is a consistent problem, I would look at hardware as being the cause of the problem. Just because its new hardware, is not a reason to discount it as being the cause. The only times I have seen the 4294... error was when the server shutdown unexpectedly and the priv.edb got corrupted. I use Trend, and I don't have circular logging on and I have never run into this particular issue other than mentioned above. I would look at disk controller or memory errors. What is the extent of your diagnostic logging for the Information Store? Set the System->(General, Recovery, and Table Cache) to maximum, and the Private->(General, Background Cleanup, and IS/DS Synch) to maximum. Monitor until the next failure, and then diagnose the cause. Be sure enable some sort of hardware monitoring and alerts for the server itself. Sorry, but that is all I think of with out combing through your log files to see try and determine the cause. Oh yeah, increase the size of your application log to at least 20MB (20480 on the size window). Clear the log and start monitoring. Let me know if I can help out in any way. I have done a number of disaster recovery restores (that was when the size of our priv.edb was in the 8-10GB range). The priv.edb is now in the 20-30GB range and only getting larger. In fact, by years end I will have to manage 2 priv.edb's about that size.

PS. If you haven't applied SP4 go ahead an apply it. I haven't run into any problems with it as of yet. Installed on 6 servers for at least 5 months.

by mckaytech

Wow, this is a puzzler! Normally, I associate this sort of situation with an impending failure of a RAID controller or hard drive, but you report it happened before it was migrated to new hardware as well as afterwards. If there are no common hardware components, that would suggest either software or an influence external to the server (power fluctuations, a virus, etc.)

You said there is plenty of room for the database - what about the space for the logs? Are they truncating properly after the backups (since you have circular logging disabled)?

If I had to make a wild guess from the information you've provided, I'd probably be looking first for a power-related problem. But other things to consider might be things like defragging software or other low-level utilities running on the server.

Then I would probably look at the date and time of each occurrence and see if that brings anything to mind (scheduled processes like backups, thunderstorms in the area, use of new equipment in the proximity of the server, etc.).

paul

This discussion is closed to new replies.