cancel
Showing results for 
Search instead for 
Did you mean: 

Proper Exchange 2003 restore of corrupted database - strange scenario(?)

BH
Level 3

Hey all -

Backup Exec 12.5 on 2003 SBS.

We started having a bad block on one disk in the RAID 1 array, this led to corrupted file system and in turn, corrupted Exchange DB.
As we were awaiting replacment disks, we kept running chkdsks to repair the underlying file system and then running eseutil integrity checks.  Luckily there was no corruption in the database the first couple of times, but then an error came up and the eseutil seemed to be doing more harm then good.  So, the solution was to restore from backup.

I have a separate snapshot backup job setup for my Exchange DB.  I run a full with verify weekends, and then an incremental throughout the week.  We had to stop the backups several nights because of the attampted repairs on the disk (to stop conflicts).

Because of the Exchange DB corruption, the last FULL exchange backup failed with a consistency check - although it appeared to backup the full database.  So, let's just say i had the following backups.  If today was 3/20 it looked something like this:

3/19 - Exchange Incremental
3/18 - Exchange Incremental
3/17 - Exchange Full (FAILED consistency check)
3/16 - Exchange Incremental
3/15 - Exchange Incremental
3/14 - Exchange Incremental
3/13 - Exchange Full (Successful)

So normally i would choose restore the last Full backup and the following incrementals.  I would choose a no-loss restore to also re-apply the existing transaction logs.  In this case, I did do that, since that is the procedure I knew.  The problem was the last Full had a failed backup status because of the consistency check.  However, the restore process worked successfully and seemed to restore the DB and all transaction files and then they replayed back and the database mounted.  However, the database FULL backups are still failing on a consistency check and the DB probably still has errors.  This was partly expected since i restored from a copy I knew failed a conistency check.

The question is, what was the proper way to do this?  Ideally I would have liked to resored the full backup from 3/13 that had no errors and then all of the following transaction logs.  However, I don't think it would have worked if I chose 3/13 and all following backups sans 3/17.  I assume that 3/17 consists of not just the full DB but also the backed up logs.  Because backup exec obscures the individual DB and log files, I can only choose to restore the IS as a whole.

If I wanted to try this again, or so I know from next time - how would I restore the good DB from 3/13 and then all the transaction logs (incrementals) from then on?

If someone can tell me not just how, but give me the theory behind how Backup Exec is handling it would be much appreciated.

thanks!
2 REPLIES 2

BH
Level 3
Still having issues here - since the restored database was based on the "corrupted" copy - all future full backups are still failing, and the transaction logs aren't being removed, causing excess space usage.

I would REALLY appreciate some guidance from some experienced users or Symantec employee.

Thanks!

BH
Level 3
Bumping this...can a symantec employee maybe give some input here??