Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Possible Duplicate:
Move raided hard drives between identical servers, preserving data

I'm a server beginner working with a Sun x4600 and an IBM x3650. In the event that one of them dies, I want to have a plan for removing the hard drives from the dead server and hot-swapping them into the other with minimal downtime.

In a series of attempted practice runs (these servers are not in production use yet), I discovered that this is a lot harder than I had assumed it would be. The receiving raid controller doesn't autodetect new drives. To make it actually expose the new drives to the operating system always seems to require a reboot and reconfiguration in the raid controller's bios, where you have to "initialize" the new drives before it will recognize them as a raid array. This is not acceptable because it wipes precious data from the drives!

How do others handle this problem? How do you hotswap raided hard drives from a dead server with a different raid controller to a live server without destroying the data on them? Is it even possible?

So far it seems like the best I could do in that scenario is send the hard drives from the dead server to a data recovery service (because apparently raid arrays are not portable at all unless the raid controller happens to be exactly the same, so popping them into another server to recover the data won't work), have blank drives ready for another server and transfer backups onto them until I have a new server to replace the dead one. That can't be right, there must be a better way, right?

I would settle for getting a server to use a solitary hard drives (without the stupid raid controller wiping the data on it) that used to be in a raid 1 array.

share|improve this question

marked as duplicate by Shane Madden, EEAA, mfinni, Scott Pack, mdpc Feb 3 '13 at 3:13

This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.

Why did you feel the need to ask the exact question twice? You accepted my answer on your previous question. If you need more information, edit your original question. – EEAA Feb 2 '13 at 20:15
Slightly different question actually. The previous question was about server with identical raid controllers (say, two Sun x4600s) - this one is for servers with different raid controllers (IBM x3600 vs Sun x4600). – user1744516 Feb 2 '13 at 20:17
"I would settle for getting a server to use a solitary hard drives (without the stupid raid controller wiping the data on it)" <-- that stupid RAID controller when used properly will save your butt some day. Learn to deal with it - it's not going anywhere. – EEAA Feb 2 '13 at 20:48
Others don't worry about this problem. Why would the server "just die"? Leverage the support contract associated with the hardware you're using. – ewwhite Feb 2 '13 at 22:02
up vote 2 down vote accepted

Different RAID controllers? All bets are off. You may be able to get it to work with very closely-related cards of the same lineage, same firmware, etc., but it's highly unlikely that this is going to be supported by your hardware vendor.

Honestly, this is where a well-implemented and well-tested backup regimen comes into play. You don't need to muck about with swapping drives between disparate hardware - just restore onto the new machine and you're back up and running.

share|improve this answer
Thanks. What if I give up on keeping the drives raided and would settle for (as a temporary stop-gap) a single non-raided drive that used to be part of a raid1 array? Is that possible? When I tried adding a single drive as a "volume", the IBM x3650 froze with a blank screen and blinking cursor right after the raid controller finished booting. – user1744516 Feb 2 '13 at 20:36
@user1744516 - give it up. It's not going to happen, regardless of RAID type. Each controller has its one proprietary on-disk structure that is assuredly not compatible with the cards of other vendors. – EEAA Feb 2 '13 at 20:40
@user1744516 - Sorry if I'm coming across as harsh. You are admittedly a beginner, and that being the case, you are spinning your wheels, wasting time and energy on doing things the wrong way. I made mistakes like that many times when I was getting started. Do yourself a favor and don't spend another second trying to get this figured out. Instead, put your energy into a good backup system. – EEAA Feb 2 '13 at 20:44

As you have already mentioned, different RAID controllers store metadata about the RAID arrays in different and incompatible ways.

Some OS's provide software RAID. Linux, for example, has mdraid, which runs in the kernel and is very mature and stable. It is fairly easy to re-assemble an mdraid array in any recent Linux box.

Of course, you lose the benefits of your RAID controler (e.g., cache, and probably I/O speed).

Linux is an example here. I think you have similar features in many other mainstream OS's (*BSDs, modern Windows Servers, OpenSolaris, maybe IllumOS).

share|improve this answer
Thanks. Unfortunately that not an option for me because I'm using ESXi 5.1. – user1744516 Feb 2 '13 at 20:40
@user1744516 Then get you some resilient shared storage and this issue goes away. Either that or just restore backups of your VM images to a pre-installed ESXi on the other server. – EEAA Feb 2 '13 at 20:42
Yes, if both servers are hypervisors, you should definetely store all the data in a resilient (and fast) NAS. Of course, as @EEAA remarked before: do your backups right. – Pablo Montepagano Feb 2 '13 at 20:47
Users are constantly posting new data to websites in those VMs, some of which would be lost if a backup were restored (we do backups daily). What do you do if you need the last few seconds of data from still-intact drives in a dead server? – user1744516 Feb 2 '13 at 20:50
@user1744516 - oh, and servers are not stupid. They're perfectly reasonable. What you are asking them to do is what is unreasonable. Bear in mind that if you had identical servers (as you should) this would not be an issue. – EEAA Feb 2 '13 at 21:17

Not the answer you're looking for? Browse other questions tagged or ask your own question.