Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations strongm on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

Exchange 2003 and Backup Exec Disk Failure - RAID 0 4

Status
Not open for further replies.

KLewisBPM

Technical User
Jan 11, 2002
294
GB
I am very twitchy at the moment, Backup exec has been backing up the exchange environment and on 3 occassions the system has crashed with one of the HDD's blinking orange! as soon as you pull the power and plug back in it boots up fine.

I have a DELL PE2650 with PERC 3/Di inbuilt. There are 5 x 33.9Gb drives in a RAID 0 configuration with total capacity 169.5Gb as drive C:

I cannot believe it is set up like this! (We inherited it)

Only 26gb is being used, it is running Exchange Server and is a Domain Member, DNS Server, Backup Server.

Is there anyway that this RAID 0 conversion can be converted into RAID 5 and retain the data?

I also have a Dell Powervault attached via a PCI PERC 3/d it has 4 73gb HDD's in a RAID 5 configuration. there are still 6 slots left and i have a couple of spare 73gb HDDs.

Perhaps i could make a new Container within the Powervault with these new HDDs and use that as a bootable partition?

Any ideas are Welcome!

Kind regards

KLewisBPM
 
Also, I think that when you move mailbox you'll get the same error.

Have you tried to move the store to the PV?
Have you tried to exmerge the users in that storage group? exmerge is great as it only takes non corrupt data...
 
Yes I installed the server into the same domain! and Exchange into the same site!

I am very reluctant to try and move the store as you can imagine!! Which is why I tried doing a file level backup! If i can isolate the file that is causing the problem then maybe I can do something to repair it? then Try to back up the server and then move the store!

I haven't tried to exmerge, what is it?

Thanks

KLewisBPM
 
So if i read between the line of your previous post. Do I need to install a server and Exchange into a new Domain? As I cannot do anything with the install i have performed already



KLewisBPM
 
If you were to remove your existing exchange server, install your replacement into the same domain, create the mailboxes for each user and then import the pst files into the new mailboxes. I'm not sure of the specific steps as i've never been in that situation but someone above will no doubt have some good tips.

You'll still be syuck with your DC on dodgy hardware though. You should move this to different hardware preferably or set up a virtual DC, transfer the FSMO roles and then rebuild your original DC.
 
Luckily its not the Main DC.. its a Member DC.

If I want to test this right through do I need to create a new Domain etc for the virtual server/exchange to operate properly?

Thanks

KLewisBPM
 
You mean member server - there's no such thing as a member DC!

If you open up the new Exchange system manager, do you see the same site name? Are the cases the same on both servers? Are they on the same network segment? Same subnet? Can you ping one from the other? Was there any issue doing domain join?

You seem to have a corruption of the priv.edb or priv.stm files that make up the first store in the first storage group so any kind of Exchange backup will be a problem. So will exmerge.

When was your last good backup and where are the transaction logs? You could restore the database on the powervault and play forward the logs and then at least Exchange is safe (ish).
 
Member server... that's what i meant :)

When I open up exchange manager on the new server installation I only see the current server. i do not see the new one!

I am going to try and back up the files 1 at a time to see if I can isolate exactly which file it is today...

i'll post back when i know.

Cheers

KLewisBPM
 
I mean file level... If I can determine which file is causing the problem i have a better chance of working out whether it will be advisable to run disk tools to recover and block bad sectors...



KLewisBPM
 
I have now found the one single file that causes the Drive to Fall over.

It is.. priv1.stm



KLewisBPM
 
look, it's the stress of the high IO level that is likely causing a soft error. What you need to do is slow it down for the backup. Try an older backapi type streaming backup to a slow media like tape.

If that doesn't do it for you, and you really have a hard error, then get yourself an external usb disk and a copy of ontrack powercontrols. The third party tool will extract what can be extracted even if the source database is damaged.

Once you actually get a backup of the mail, you heed to think about AD and Exchange. If the exchange server is not also the domain controller, then it's a piece of cake; wipe the box, reconfigure the disk, install the OS, then do a disasterrecovery install of exchange. Once you're up with empty mailboxes, use powercontrols to inport the mail you exported to the usb drive.

If the exchange server is also the domain controller, you need an extra server. Install windows, make it a domain controller, make it a GC, then transfer all the FSMO roles to it. At that point, you can demote the original server (unsupported but exchange is already broken) then wipe the original exchange server and reload/disasterrecovery install as above. You demote to avoid the messy details of just killing the server and doing metadata cleanup.



 
I have been doing a bit of reading on the STM file and have come across an interesting way of reducing the file by running an offline defrag.

Based on the information throughout this post does anyone know if the defrag process would do more damage than its worth?

KLewisBPM
 
The defrag process copies the databases to a temp location and then defrags it. If you kill the defrag process then the original is still intact. That's the thoery anyway.

Have you tried exmerging out the mailboxes?
 
No not yet. So Far I have Discovered which single File is laying on the "Disk Error". I have created A virtual environment within the same domain with Exchange however this is not correct. I need to create a Server environment in a different Domain so I can use it as a test restore server.

I want to be able to prove that when i take the Main Server down I will be able to bring it back up successfully!

xmsre's advice regarding the slowing down of the back up intrigues me.

What amazes me the most is this server just keeps running and running, its been like it for nearly 3 months. It only falls over when you want to run a back up of First Storage Group or do a file level backup of the priv1.STM

If there was a way to make a copy of this particular file and move it to a different Redundant Array... Life would be so easy!

Another Question. Are the backed up mailboxes no good at all with out the First Storage Group?

Cheers

KLewisBPM
 
Is Exchange standard or enterprise edition? if it's enterprise, create a new storage group on different disks and move the mailboxes across to the new storage group one at a time. If you have installed a second exchange server but that's not working right, i think you should concentrate on that first.
 
Interestingly I weren't 100% sure whether is was standard or enterprise so I tried to add a second storage group and it allowed me to! I pointed it to a redundant array drive to create its new edb and stm files.

Perhaps this could be the way Forward? We only have around 45 Email users... If I move all the users' mailboxes will they continue to work seamlessly? If i move all users, I could remove the First Storage Group and then run a proper disk check on the dodgy drive?

Cheers

KLewisBPM
 
If you create a new store (in a new storage group or not), you can then movemailbox all of the user mailboxes to the new store. All items are fully promoted to the edb, so your new stm will be minimal. Once the mail is moved, you're left with the system attendent and smtp mailboxes only on the first store. Dismount the store and delete the edb and stm files. Now, mount the store and click through the warning that the files are missing and a new store will be created. Once the new store is created and mounted, the system attendant and smtp mailboxes will automagically be recreated. Presumably, you would then move this new minimal store to another storage device.

This still doesn't address the issue that it's one big C: drive with RAID 0. You'll still probably need to take the server down, reconfigure the disk, do a disaster recovery install, and reattach to the databases.



 
Definately worth a shot, what can you lose?
I think the user has to come out of outlook for the transfer. Last time I did it was during working hours and i think the user got a message to close outlook and reopen. Their mailbox was disconnected during the move. Try it with your admin one first and see what happens.

If you can move the store to the other disks, i would only see this as a temporary move before rebuilding that server. It will at least allow you to get a proper backup of exchange.
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top