Safe to offline a disk in a RAID-5?

[Nick_Parrott at Dell.com]

Hi Mark,

Just for peace of mind, the official Dell answer;

A drive needs to be "prepared for removal" before swapping it out with a replacement. This is sometimes referred to as "prepare to remove" or "offline" in OMSA. If the disk has failed and rebuilt already, I'd definitely suggest doing one of the above commands before yanking the disk. If the disk is already "failed" or "offline" then it's safer to pull it, but still try the "prepare to remove" if possible.

I've had controllers before now (purely down the nature of SCSI) fail entire arrays when a good *or* bad disk is tugged. It's your call, but we prefer you to prepare the controller for the bad news, just in case..

Incidentally, when a disk fails, I'd advise contacting Dell before rebuilding. Every disk fails for a reason, if you give me the controller log, I'll tell you why. I don't like the "rebuild the disk and see what happens" plan, unless I know what the issue is (from the controller log) and know the rebuild is required to resolve the overall issue.

Good luck

-----Original Message-----
From: linux-poweredge-bounces at dell.com [mailto:linux-poweredge-bounces at dell.com] On Behalf Of Mark Watts
Sent: 23 May 2008 15:29
To: linux-poweredge-Lists
Subject: Re: Safe to offline a disk in a RAID-5?


On Friday 23 May 2008 15:18:03 Zembower, Kevin wrote:
> I had a drive fail in my PE 2850 with PERC 4e/Di. Removing and 
> reseating the drive caused it to rebuild successfully. However, to be 
> on the safe side, Dell shipped me a replacement.

I've done this several times with 2550's

> I was just preparing to remove the failed drive from my RAID-5 array 
> using the OMSA Web GUI. Instead of just yanking it out of the chassis, 
> which theoretically shouldn't cause any harm, I want to 'Offline' it 
> in OMSA and then remove it. When I went to do this, I got the warning,
> "Warning: Making a physical disk offline may result in data loss. Are 
> you sure you want to offline this physical disk?"
>
> Is it safe to continue to offline this disk? If not, what is the 
> preferred procedure to replace a disk in a RAID-5 array without taking 
> the host off-line?

Yanking the broken disk and replacing it with a new one is kinda what "Auto-rebuild" is for; I've certainly never had any issues with doing that but YMMV.

I suspect that OMSA isn't quite clever enough to work out that the disk you want to remove is part of an otherwise healthy RAID-5, so gives you the warning anyway.

Mark.

--
Mark Watts BSc RHCE MBCS
Senior Systems Engineer
QinetiQ Applied Technologies
GPG Key: http://keyserver.veridis.com:11371/search?q=0x455420ED