Exchange server disaster recovery - no information store!

[superatticman]

Good Evening

 

We recently had a catastrophic hardware failure of an exchange server on a relatively small domain based network. Bascially we lost 3 scsi hard disks in quick succession meaning that a DR from backup was the only solution onto a spare server with identical hardware configuration.

 

Configuration:

 

Windows 2003 Server SP2

Microsoft Exchange server 2003 (latest sp)

Arcserve 12 sp2 Backup server

 

Managed to do the DR 'by the book'. And now have a restored server.

 

However, we have noticed the information store service is not starting, and a quick examination reveals that the .edb files are missing. We have also noticed that the .edb files are missing from the previous backups. We know why as the .edb was moved to a different volume, but the change not made in the backup to ensure the information store on the new volume was backed up too!

 

However, we DO have recent brick level backups of the individual users mailboxes. Obviously we cannot restore these, as there is no information store to restore to!

 

Any suggestions of how we move forward with this?

 

Just some ideas:

 

a) Create a brand new info store and restore the brick level back to these?

 

b) Restore a 6 month old version of the information store of which we should have on archived backup (before it was moved to another volume), and then restore users folders on an individual brick level basis of which would bring them up to date.

 

Thanks for any advice with this.

[sukh]

Hi

 

The disks which had the EDB files, are they still accessible? Or are all 3 disk knackered?

 

There are tools out there whereby you can recover the mail data if you still have access to the EDB files.

 

Have you started the recovery procedure yet or not?

 

I would create a new DB and roll forward, I never roll back a transaction. Be careful if you decide to import he bricklevel PST files, not to create duplicates, this can be scripted using Exmerge.

 

Let me know how you get along and if you need help.

 

Cheers

S

[superatticman]

Hiya

 

The disaster recovery is now completed.

 

Decided to mount a new information store, and then restore the brick level backups. Considering there is only 20 mailboxes on this server, manual restoration of each mailbox was fine. Managed not to create duplicates too.

 

Regarding the disks, I've never known a server with 4 Raid 5 HDD's and an online spare to simply 'blow out' like it did. There must have been some kind of major failure on the HDD controller which in turn damaged the disks.

 

Whats important now is that the backup is set correctly to back both brick level and database level!

 

Thanks for your input, is much appreciated.