Sign in with
Sign up | Sign in
Your question

Replacing Server Hard Drive

Tags:
  • Support
  • Hard Drives
  • Business Computing
  • Servers
Last response: in Business Computing
Share
April 21, 2014 10:55:03 AM

Hello.

Recently took over support on a SBS 2011 server which is the sole DC in the organisation, along with hosting the usual MS Exchange services and acting as the main file server.

It's running on Western Digital Green hard drives. They are SATA and less than 7200RPM due to being power saving. The first thing I'd like to change is these drives, most likely for some 10k VelociRaptors as it doesn't have any SAS connections.

Getting to the problem - I will struggle for permission to take the server offline for any more than an hour. I doubt this will be long enough to clone ~500GB of data. There is no RAID setup either so I can't swap out the drives one by one and let them rebuild.

Looking for recommendations on how to approach this. I imagine some imaging software could be the solution?

TL;DR
I want to clone a servers hard drive without taking it offline for any more than 60 mins.

More about : replacing server hard drive

April 21, 2014 11:05:35 AM

60 minutes, with 100% certainty?
Not gonna happen.

A server (DC and Exchange) running on WD Green drives? That is a recipe for fail.

DriveImageXML can maybe, probably, create an image from a running system. I've done it in a home environment, but never with a server.
m
0
l
April 21, 2014 11:20:32 AM

USAFRet said:

A server (DC and Exchange) running on WD Green drives? That is a recipe for fail.


Precisely why I want to change them asap! Performance is surprisingly not too bad but I simply don't trust them in a server.
m
0
l
Related resources
April 21, 2014 11:22:57 AM

This is what failover is for. Tell them that their poor planning has led to a situation where they will have significant downtime or risk much more serious downtime and/or data loss further down the line.

Obviously, the politics of this is bad all around, so I'm not sure I can help you there.
m
0
l
April 21, 2014 11:42:48 AM

If push comes to shove they will be told precisely that - accept the downtime or accept that I hold no responsibility for any future performance issues or data loss.

But naturally I would rather it never came to that.
m
0
l
April 21, 2014 11:48:35 AM

Swapping the entire infrastructure to new drives, within 60 minutes, and not having a tested plan and tools to do it.....can't be done.
m
0
l
April 21, 2014 12:12:28 PM

I'd have to agree with USAFret, there is no way this could be done within 60 minutes. You did mention that the server is running a multiple WD hdds, and since there is no RAID, is one hd running the os while the other is storing the company data? Also an imaging software I use is clonzilla, I use it all the time on desktop to take a complete image of the computer. I have never tried it server side, but since there is no RAID that might help you out.

My recommendation would be to try after hours take an image of the hard drives replace with the new hard drives, and restore the image onto the new drives. Doing it after hours will allow for any trouble shooting and if nothing works you will always have the old drives to fall onto.
m
0
l
April 21, 2014 12:30:42 PM

I've used CloneZilla on servers before, but I don't believe it can be done with running drives.

Overall, this is the type of thing you try 2 or 3 times before you go 'live'.
2AM messing around in the server room over a couple of nights. Try it once. Did everything work perfectly? If yes, try it again the next night. If it still works, then maybe you can go live with the swapover.
m
0
l
April 21, 2014 12:31:22 PM

Yes one drive is used for the OS and a few other things - roughly 120GB of data on that.
The other drive is used for data - roughly 350GB of data on that.

I'm not so concerned about the data drive. That can be copied at any time by just stopping a few services.


I've used Clonezilla on a test server before. It works at block level so theoretically it can clone any OS. However as USAFRet just said, it requires the disk to be offline before it can be cloned.
It would be my first choice given more downtime.


I think an overnight shift is looking very likely at this point.
m
0
l
April 21, 2014 2:19:35 PM

there is a solution for your problem.
you need make backup of your AD , EXCHANGE , FILES.
in this scenario your current server be called server A.
1. make a current backup of all your data on server A (files , exchange , active directory).
2. you install server B with SBS 2011.
3. get server B into domain install necessary roles : AD , EXCHANGE , DNS , FILE SERVICES.
4. move roles from server A to server B (AD (fsmo roles) , EXCHANGE mailboxes , FILES) (this step your have to plan very carefully and check guides idk sbs never worked with it so make your reading , if 2 sbs 2011 cant coexists you can migrate server A to server B.
5. remove exchange from server A.
6. demote server A AD's. (this done after you check there is no replication issues between servers !!!).
7. take server A offline and change hardware (disks , recommend also good raid controller).
8. install server A OS and test HW.
9. from here you go to step 3 and reverse roles.

some other issues you have to plan :
1. the exchange owa , smtp services setup in firewall . change them temporally to server B.
2. dont forget to backup and restore exchange certificate from server A.
3. copy files from server A to B will take 3-5 hours which files wont be accessible (maybe just for reading).
4. this process will take over 10 hours i suggest do it over weekend

plan your steps put it to writing , make backup and good luck.
m
0
l
April 21, 2014 5:25:13 PM

Implementing a major hardware change like this is something that can't feasibly be done while the server is running. There will be downtime, one way or another. Even cloning the data will not be feasible during operational hours.

If you want to have this done with minimal down time, the only option I really can see is converting your current server to a virtual machine and move it to separate hardware, which isn't a bad idea anyways. Just replacing out hard drives, configuring your necessary RAID arrays (if any) and transferring data and settings is going to take a deal of time. And always plan for more time than minimum, because truth be told I've run into several situations like this where something like the hard drives not cloning properly caused three times longer down time than it might have taken.

I would present it to them like this, perhaps. Option #1 is they must take their server down for the good part of a day while you migrate data from the old hard drives to new hard drives and ensure everything is in operation before coming back online. Option #2 is they get a new server with the new hard drives you are looking at, and migrate everything from their current server to the new server with less downtime (but still at least some, perhaps an hour or two.)
m
0
l
April 29, 2014 3:11:49 PM

Got this completed the other night with a different approach.

Purchased "EaseUS Todo Backup Advanced Server" as there were always future plans to improve the backup routine, we just pushed them forward. Dropped it straight in at the deep end with a full system restore to dissimilar hardware!

- Scheduled a full backup during the day
- Topped this up a with a differential backup right before shutting down
- Swapped over the drives for new ones (straight swap, no RAID)
- Booted into the EaseUS recovery environment and restored to the new drives

Worked incredibly well. Roughly 4 hours downtime including some testing but they now have faster, more reliable drives and a solid backup in place.


Thanks for the help and for preventing me attempting this in a 1 hour window. That would have been messy.
m
0
l
!