ARGH - Backup Exec 9.1 and Exchange 2003

[Zoplax]

Hi all, I've followed the guide at Veritas' site for backing up Exchange.

http://seer.support.veritas.com/docs/266074.htm

I've set my Windows 2003 Standard server up so that only the Microsoft Information Store is backed up, not the shadow copy components.

I've deselected files and directories specified in Veritas' document which the Exchange Agent already takes care of.

I've also excluded all Exchange data directories from the virus scans (Symantec Antivirus Corporate version 8).

Despite this, after a recent reinstall of Backup Exec 9.1 and its latest service pack (3), I still get the following errors during Backup:


Backup - \\myserver\Microsoft Information Store\First Storage Group
WARNING: "\\myserver\Microsoft Information Store\First Storage Group\Mailbox Store (myserver)" is a corrupt file.
This file cannot verify.
WARNING: "\\myserver\Microsoft Information Store\First Storage Group\Public Folder Store (myserver)" is a corrupt file.
This file cannot verify.
WARNING: "\\myserver\Microsoft Information Store\First Storage Group\Log files" is a corrupt file.
This file cannot verify.


What the heck else could be locking the Exchange resources?!? >:-< I've run out of ideas.

Thanks for any help.

 

[steveot]

Take a look in the event viewer in the APP log and look for the source being ESE during the time of backup. You will most likely see the root of your problem.You exchange database and logs are corrupt and ESE is returning the results to backup exec. You need to contact microsoft.

 

[secureops]

Corrupt files with an exchange job can mean only that one item is locked or in a state that cannot be verifed. I get these erros quite a bit and it is usually related to data that is in use or a permissions issue with a specific resources. I have also cleared up this problem by running a defrag & compact (eseutil) on the database -but this is only in cases where I cannot access the data causing the problem.

 

[Zoplax]

Ok, I filtered the Application event log for messages from ESE and all I've found are informational messages about Exchange database defrags being successful, shadow copy freezes started / stopped, errors about backups being stopped due to (my) manual intervention, and a few warnings from months ago about the database taking unusually long to read from a particular resource (but these haven't recurred since).

 

[steveot]

Have you tried running a NTBACKUP to disk on the IS with detailed logging turned on in NTBACKUP?

 

[Zoplax]

No I haven't, I'll give that a try.

 

[Zoplax]

Ok, I tried running a backup with NTBACKUP but it failed. I got an error at first saying "Backup is still active", and found an article where it suggested first backing up the System State so that it would clear whatever flags first, and then try the Microsoft Information Store separately in a subsequent backup. This didn't work.

I'm thinking now I may just take the server down and use ESEUTIL to repair and defragment the database, it seems like there may be corruption which is messing up the database.

 

[Zoplax]

I took down the server and ran ESEUTIL against the private information store.

It started out as a 12 GB database, and took about an hour to defragment (I chose the option to build a new database as well) on a Dell PowerEdge 2600 with dual Xeon 3 GHz CPUs, 1 GB RAM, and SCSI hard drives in RAID 5.

Surprisingly it ended up just over 4 GB after the defrag and compact operations were done! Luckily though Exchange started up smoothly, and when I performed the next Full backup in Backup Exec, the Information Store errors vanished.

So, this appears to have worked!

 

[secureops]

Glad to hear it.