Link to home
Start Free TrialLog in
Avatar of Zoldy2000
Zoldy2000Flag for Canada

asked on

Exchange 2003 backups failing

We are running Exchange 2003 and using Veritas Backup Exec.

The backups on the Exchange server have been failing.  The error is database or database element is corrupt.

I have noticed that there are a bunch of E0031***.log files within MDBDATA that constantly being modified.  
Is this normal and are these two things  related?
ASKER CERTIFIED SOLUTION
Avatar of Sembee
Sembee
Flag of United Kingdom of Great Britain and Northern Ireland image

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
Avatar of Exchange_Admin
Exchange_Admin

In your backup are you backing up the Exchange mailboxes or the information store?

It sounds like you are backing up the mailboxes. This is also called a brick level backup (BLB).
When I configure a backup I only backup the Information Store. When the Information Store is backed up then the log files you mention will be flushed.
When doing a BLB backup the logfiles are not flushed.

If you properly configure your deleted item retention time then you will greatly minimize the need to restore individual messages, etc. (which is the main reason for doing BLB's).
If in fact you do have to restore something that has expired the deleted item retention time then you can use the recovery storage group feature of Exchange 2003.
Deleted item retention and recovery storage group information can be found by checking the MS Knowledge Base.

Hope this helps.
Russ
Avatar of Zoldy2000

ASKER

We are backing up all the mailboxes. I will be  running a full backup tonight.  Should I not run a block level backup?
There is nothing wrong with brick level backups apart from being slow, useless for disaster recovery and usually a waste of time and space. They backup just the data, so you would lose all of the settings in the mailboxes.
A brick level backup is also not considered a backup of Exchange, so doesn't flush the transaction logs.

What most people (myself included) recommend is that brick level backups are only carried out if you have the time and space after a full information store backup. They should not be done instead of an information store backup. If the backup starts to overrun in space or time then they would be the first thing to drop as they are in effect a duplication of the information store backup.

Simon.
Ok, you suggest running ntbackup.   How long does this take? The exchange data is around 30 Gb.
Should I run the ntbackup before, after or in place of the full backup... or does this not matter.
If you own Veritas with Exchange licenses then you don't need to use ntbackup. The only reason that ntbackup is suggest is when Veritas doesn't appear to work correctly. Ntbackup is pretty basic but does the job.

Simon.
Ok, there is a file that appears to be corrupt.  I have searched for this file within the users outlook, but was unable to find it. It must be located somewhere on the Exchange Server.  I did a search on the Exchange but again couldn't find it. Where are these messages kept?
If you are getting corrupt messages from a mailbox then you are not doing an information store backup - you are still doing mailbox level backups.

Simon.
I am checked the job setup.  Both the mailboxes and information store are checked to be backed up.
Select just the information store and see what happens.

Simon.
Was able to find the corrupt file.  Once deleted backups were successful.