Nick

Hi, can anyone help with this?

RAID5 with 5 disks + 1 hotspare.

I noticed last week that disk 0 of the array was displaying a red light, although the hard disk activity light was flashing. I decided to reboot to check the RAID utility program & on reboot the red light went out & the utility program showed no problems. I had been under the impression that RAID5 could repair itself on the fly? If so, why the red light?

Since that, I've been watching the hard disks lights regularly. The disk activity light on drive 0 does not seem to flash in synch with the other disks. Is this normal? Maybe an impending failure?

HHxx

Raid 5 doesn't repair itself on the fly as such.

The raid set can lose 1 disk and be still be functional. The repair part is where it can rebuild the missing disk's data onto a replacement drive or the spare from the data and the parity info on the remaining drives.

Which manufacturers raid card are you using?

There is usually a log for errors? Maybe statistics on the drive, could have been excessive timeouts?

Did the spare kick in? I would be worried about that more and replace the potentially faulty disk.

Nick

Thanks for your reply.

The server is a Fujitsu Siemens TX300, the RAID is onboard. I don't think that there's a controller log or stats, but I'll look for one. The hot spare is not active & is still designated a spare in the RAID BIOS.

I think that I agree that disk 0 should be replaced. Do we just pull it out & pop in the new one, then run a rebuild on the new disk?

druddle

I would treat the red light as a warning and replace the disk asap. As above, if you lose the disk the RAID set will carry on running and rebuild the data that is striped across the remaining disks.

The disk may not have failed, just had an error which is why the hot spare may not have been used.

There may be a log somewhere, i will see what I can find out for you. Can you tell me what the RAID controller is (from BIOS) ? I might work for FSC......(dont specalise in Intel/AMD servers but if i can help I will).

Dave

Nick

Thanks very much Dave!

I think the controller is a MegaRAID SCSI 320-2E. ServerView & GAM are not installed. I found a SCSI RAID user guide (for the RX600) online & it looks as though the disk 0 should be treated as a "disk likely to fail". So bearing in mind there is already a hot spare installed the steps appear to be:

Buy new disk
Boot to SCSI BIOS
Remove Disk 0
Hot spare will start building automatically (maybe)
Insert new disk in disk 0's slot
New disk 0 will become hot spare

I'm not sure if there's a way I can make the new disk a straight replacement for disk 0 unless I deallocate the hot spare first?

druddle

Should have installed the OS with Serverview as it automatically installs all the correct drivers and firmware. Mind you, you are not the only person who hasnt !!!!

That looks Ok to me, the only way to get disk0 back to be used and the old hot spare to be the hot spare is to wait until the RAIDset has been fully rebuilt and give the disk a couple of days to make sure its not a DOA, then pull the disk. It will build disk0 in as a data disk and when its done, put the hot spare back in. DONT REMOVE THE HOT SPARE FIRST as when you then swap disk0 you will have lost 2 disks in your RAID5 set and your data will be destroyed. Only RAID-DP (6) can handle 2 disks failing in a RAID set. Voila ! If it all goes Pete Tong dont tell the guys in support i said anything

Dave

Nick

Thanks for your help Dave!

I've never used ServerView because it has not had the correct drivers in the past. So I just spend several hours hunting through the 6 CD's for the correct drivers! Once bitten....

What do you think though of my suggestion of de-allocating the Hot Spare in BIOS first? AFAIK, the Hot Spare is not used in the array if it's running. Then I can move the Hot Spare to the Disk0 bay & let it build into the array & put the new disk into the Disk5 bay so it looks all nice & tidy.

Buy new disk
Boot to SCSI BIOS
Deallocate Hot Spare Disk 5
Remove Disk 0
Insert old Hot Spare in disk 0's slot
Wait for build to complete
Insert new disk & allocate as Hot Spare.

Is that ok, or just unnecessarily complicated?

Or I can just leave the new Disk0 as a hot spare - so I don't push my luck!

druddle

I would just make one change at a time so if things go wrong you know whats caused it. In your plan you are removing the hot spare and removing disk0 at the same time. Could result in something unexpected happening (god forbid you lose another hard disk....)

Dave

Nick

OK, good advice thanks!

Buy new disk
Boot to SCSI BIOS
Remove Disk 0
Wait for build to enable Hot Spare into array
Insert new disk & allocate as Hot Spare.

?

druddle

Yep thats good. Once its all rebuilt and has been running with new disk for a few days you could pull the old hot spare and that would rebuild the new disk 0 as part of the RAID set. Once thats done you can add the old hot spare back in as the hot spare.

Dave

Nick

Will do.

I appreciate your advice.