Hot-plug drive in DL360 freezes Windows 2003

[Franc van de Westelaken_1]

Hi,

Our DL360G3 is configured for RAID1+0 using 2 36Gb drives. When one drives failes during operation and I hotplug a new drive while running Windows 2003, the OS completely freezes, only the mouse pointer works but nothing else. Only turning the server off and on again helps. After that, the bios of the 5i raid controller (version 2.56) recognizes the situation and aks if it must rebuild the mirror. Likewise when removing a drive while windows 2003 is running results in an OS freezes. After a reset, the beos recognizes the situation again and asks for interim recovery and then boots the OS. Why doesn't this work while running windows 2003 ? This way the mirror configuration is of no use, since continues operation of the server is not guaranteed since you have to hard reset the server. I've installed the server from Smartstart 6.40.

Franc.

[Sunil Jerath]

Hello Franc,
Even though the hard drives are hot pluggable I always recommend bringing the server down and then replace the failed drive. It's just out of the habit and a similar experience likes yours that I suggest replacing the HDD when the server is off. In RAID 5 I have not experienced the same situation though. I hope this answers your dilema.

Regards,

[Michael Schulte zur Sur]

Hi,

Raid1+0 with 2 drives? I can't imagine that. How do you do the raid? Hardware, Software?

greetings,

Michael

[Franc van de Westelaken_1]

Hello, ok I understand. Problem however is, that when the drive fails, the complete OS freezes and I can't shutdown the OS gracefully and have to turn powercycle the server. However, I simulated the failure by pulling the drive out. Is this situation different than when a real failure occurs? In other words, when the drive fails for real will the system be able to recover from that easier then when pulling a drive out ?

Franc.

[Franc van de Westelaken_1]

Hi Michael,

my first reaction was the same as yours. But yes it's possible. The raid configuration utility reports the array as a RAID 0+1. From the help file of the ACU:

----------------

RAID stands for Redundant Array of Inexpensive Disks. RAID 1+0 (drive mirroring) is a fault tolerance method that uses 50 percent of drive storage capacity to provide greater data reliability by storing a duplicate of all user data. Half the physical drives in the array are duplicated or "mirrored" by the other half.

RAID 1+0 first stripes your data across half of the disks, then mirrors this data to the other half.

Drive mirroring creates fault tolerance by storing two sets of duplicate data on a pair of disk drives. There must be an even number of drives for RAID 1+0. This is the most costly fault tolerance method.

If a drive fails, the mirror drive provides a backup copy of the files and normal system operations are not interrupted. The mirroring feature requires a minimum of two drives, and in a multiple drive configuration (four or more drives), mirroring can withstand multiple simultaneous drive failures as long as the failed drives are not mirrored to each other.

---------------

I never heard of it before, but it seems possible.

Franc.

[Sunil Jerath]

Hello Franc,
Just a word of caution do not unplug the working HDD from the system while it's up and running. I can quote you tons of examples as to what happened when some other clients pulled the good HDD out of the system. I'm surprised you did not end up building the OS altogether. The HDDs are hot pluggable and NOT hot unpluggable. If the system froze when you removed the live HDD then it's normal that you experience the OS halting.

Regards,

[Michael Schulte zur Sur]

Hi,

finally I got it. You make a cross with the data. When you removed the disk, was the disk actually marked as bad or still running? Controllers don't like it, when you abruptly remove a disk, which is under load. Once I caused a reset with this on a machine with hundreds of user on it. :-(

Michael

[Franc van de Westelaken_1]

Hi Sunil,

thanks for your clarification. Guess I was under the assumption that you can pull the drive while it's running.
But is there a safe way to fail a drive and simulate the recovery process ? Is there some sort of utility to do it ? I need to test the entire process before I can take the server into production.

Franc.

[Sunil Jerath · Accepted solution]

Hello Franc,
You are not the only one thinking that you can pull the HDD from a functioning server when it has not failed. The word Hot Pluggable has a lot to say about it. But it really hits hard when we pull the drive out when it has not failed. You can short circuit the back plane board as well. I am letting you know as per my personal experience I had with lots of clients in these unfortunate situations. As for as I know we do not have any simulation or utility to just fail the HDD and then see the behaviour of the server. Sorry but this is the bitter truth.

Regards,

[Franc van de Westelaken_1]

Hi Sunil,

ok, thanks for your prompt answers. Indeed most people think you can plug and unplug the drives while it's running only because of the word 'hot pluggable'. I've asked several people and they all say, sure it should work...

About the simulation, well I have to take HP's word then that the controller functions as expected (but that shouldn't be too difficult) and we'll see what happens when the moment is there (if we ever get to that moment).

Again thanks.

Franc.