Nightly Exchange Crash

[royalmail]

At the moment an Exchange server I manage crashes every night around 4.00am. It is 5.5 running on 2000 Server. Whatever happens causes a blue screen. The last entry in the Event Viewer is:

ESE97 Event ID 179
The Microsoft Exchange Directory Service (1088) online defragmentation is beginning a full pass on DSADATA\dir.edb

This online defrag would appear to crash the server each and every time it runs.

Anyone know how I can prevent this? I have run Performance Optimiser again, and just now reinstalled SP4, but I can't say I am optimistic about that fixing anything..

 

[Taffyboyo]

It looks like you may have a corrupt Exchange directory.

Try doing an off-line defrag on dir.edb using eseutil.exe from the commmand prompt.

You will need to stop all the email services first.

 

[dmandell]

I think taffyboyo has the right idea. Remember that running isinteg can take an extended amount of time depending on the size of the exch. databases. Be prepared to possibly take users offline for many hours.
Dana

 

[royalmail]

Thanks for the help.

The issue is definitely with the Directory Store though, and not the Private or Public information stores. I defragged the DSA database offline yesterday and it completed without issue. Regardless, the machine crashed this morning at midnight. Last thing in the Event Viewer is, 'Event ID 179 - ESE97. MSExchange DS (976) Online defragmentation is beginning a full pass on database \exchsrvr\DSADATA\dir.edb.'. Pow!

 

[dmandell]

Did you already try running the IS/DS consistency adjuster too?

Also, I suppose you can temporarily stop the online defrag until you have more time to research the problem. -Dana

 

[blackrabbit]

Did this issue ever get fixed? I'm having the exact same thing hapen to my exchange server and we can't figure out how to stop it.

 

[royalmail]

Well, I resolved the issue, but I didn't fix it.

This probably isn't what you want to hear, but nothing I tried to fix the problem worked. I did disable the online defrag function, which made the server more stable, but that is only really a short-term fix as you know.

I was fortunate in that the Exchange server was being used as a routing server rather than a mailbox store so I didn't have as much pressure - I could come up with a work around that didn't affect the users.

Anyway, I built up another Exchange server and added it to that Site. Moved everything over to the new box, followed all the steps for how to remove the first server in a Site, then killed the troublesome box.

I then added another box and added it to the site, meaning that the original box was gone, and two new servers had taken its place. I then dedicated one box to Directory Replication and the other to X400 routing connections. Since then the two have been working well together.

If you find a fix to it I would be interested to know.

All the best.

 

[joeg701]

This happened to me once a few years ago. I had Exchange 5.5 installed on a box with a Raid-5 hard drive controller. One of the drives had gone bad in the raid array, and every night during defrag, the store would get corrupted and the server would crash. My mistake was not installing the Raid array monitoring software - I didn't know the drive was bad until I went into the controller BIOS during reboot and looked at the status.

 

[royalmail]

In my case the server ran off a single SCSI drive, so I guess it is possible that parts of that drive had disk errors.

Next time someone has this issue, could you maybe try moving the directory data to a different partition or drive using Performane Optimizer and seeing if it makes a difference.

 

[GaviD]

I would suggest you check what other functions are running at the same time, esp WRT backups & antivirus activities.
Running a backup at the same time as a defrag could cause serious instability.
Also check if you are running NAV for MSE. Through bitter experience I have discovered that it is better to disable it while running any major maintenance.

 

[royalmail]

Just for the record, in our case the Exchange server was only a routing server, and although it was being backed up, this only took a matter of minutes as there was no big mail database. In addition, this backup didn't clash with the online defrag. Furthermore, there was no AV software installed on the mail server.

 

[blackrabbit]

I wound up turning off the onlinedefrag and it stopped it. Its only a short term fix but we have enough space right now to work with. We building a new one soo with win 2k3 and exchange 2k3 or 2k.