Exchange Information Store Failure

L

logo29a

Gawd
Joined
Feb 4, 2004
Messages
649
Hi,

We just had one of our main mail servers fail due to a missing log file. I was wondering if anyone had ever ran eseutil.exe before. We are currently in the process of repairing the database with this utility and it has been stuck on the "reparing database" portion for about 2 hours now. It is at the point where it says "deleting unicode fixup table". It hasn't moved for a while, but the .raw file is being written to every now and then and the process doesn't appear to be hung. Has anyone experienced this before?

PS ~ I understand that this utility generally takes a while to complete. I just want to make sure that I'm not wasting valuable time if it is hung up.

Thanks.
 
You are fine. eseutil.exe always takes awhile to complete. If you are writing the output to a log file it tends to run quicker
 
The times I have had to use eseutil to check a DB, or fix a DB, it took all day. Then again, our DB was up around 16GB at the time. So it takes a long time. Plan on it being down for a day.
 
The time I had to use it, it took a while to complete, and this was on a small 700MB DB or so.
 
Missing a log file? Hmm... That's not good. Have backups ready to go? I say that because ESEUTIL is the low level fix tool. It's looking at the database itself. ISINTEG is the higher level Exchange fix tool. Sometimes ESEUTIL has to do so much that Exchange can no longer mount the database.

What errors in the event log did you get when trying to mount the DB? How big is the DB? Our email database is 80GB. It would literally take days to ESEUTIL that.
 
There error logs were from the information store and stated that the last log file out of a group it needed couldn't be loaded. Where are these log files? I thought that they would be in the mdbdata folder but the names weren't the same. The log it was looking for was just a number where all of the logs in mdbdata are ee0xxxxx.log. Eseutil finally completed successfully. Now I am about to run isinteg -fix before remounting. I assume that isinteg is integral to fixing this problem and that it cannot be skipped.
 
Could anyone ballpart how long isinteg.exe -fix will take to run? Eseutil took approximately 4 hours to complete.
 
Can't really say how long..but it may be a wihle.

The log files are stored in your transaction log folder. If circular logging is enabled it will overwrite old logs...which is a bad thing when something like this happens.
 
NetJunkie said:
Can't really say how long..but it may be a wihle.

The log files are stored in your transaction log folder. If circular logging is enabled it will overwrite old logs...which is a bad thing when something like this happens.
Click to expand...

Is the Mail.log folder my transaction log? No ballpark figure? :D

It took approximately 4 hours to do the full Eseutil. Should I expect similar times for Isinteg or shorter?

Thanks for everyone's help.
 
It looks like all of this started when the server was trying to backup E00.log to another file. It failed. Then when it tried to start writting to E00.log again it couldn't and everything crapped out then.
 
logo29a said:
It looks like all of this started when the server was trying to backup E00.log to another file. It failed. Then when it tried to start writting to E00.log again it couldn't and everything crapped out then.
Click to expand...

That's exactly what happened to ours once. It's never happened again on the same hardware. We assume it was some hardware glitch. We did a tape restore to get ours back since we couldn't wait on ESEUTIL to run for 3 days.
 
For future reference use some sort of redundancy with Exchange. I would suggest 1 server per every 10Gb os storage or so. It took us approximately 12 hours to fully repair a 20Gb private store.

However, I am now an expert at information store recovery. :D Not only that but I think I got it in good with the boss due to all of my hard work.
 
logo29a said:
For future reference use some sort of redundancy with Exchange. I would suggest 1 server per every 10Gb os storage or so. It took us approximately 12 hours to fully repair a 20Gb private store.

However, I am now an expert at information store recovery. :D Not only that but I think I got it in good with the boss due to all of my hard work.
Click to expand...

One server per every 10gig?? Thats a bit overkill ya think?
 
logo29a said:
For future reference use some sort of redundancy with Exchange. I would suggest 1 server per every 10Gb os storage or so. It took us approximately 12 hours to fully repair a 20Gb private store.

However, I am now an expert at information store recovery. :D Not only that but I think I got it in good with the boss due to all of my hard work.
Click to expand...

We just moved to a failover cluster w/ SAN. Instead of 8 - 12 servers we're just going to split up the database in to multiple Stores. My suggestion is that unless you have an ongoing issue skip repair and just do a restore. It's a lot faster.
 
NetJunkie said:
We just moved to a failover cluster w/ SAN. Instead of 8 - 12 servers we're just going to split up the database in to multiple Stores. My suggestion is that unless you have an ongoing issue skip repair and just do a restore. It's a lot faster.
Click to expand...

This is the way many companies including ours do it today. Smart and super scalable.
 
Ok, so we got both of our stores back up and running. However, now mail is queuing up for one of our public folders. The weird thing is that if I send a message to it from a yahoo account, the mail goes through. If I send it from a local account, it gets stacked up in the queue. Could this be some sort of routing problem internal to our organization? What could be causing this?

Thanks.
 
You must log in or register to reply here.
Back
Top