Array 10 with online spare - disk replacement procedure

[Adam Olejniczak]

Hi,

There is one thing I would like to confirm.

Situation like this: we have Smart Array 6400 for example with RAID 10 setup and online spare.

Disk fails, online spare is being put to use and array is rebuilding (hpacucli show something like: logicaldrive 1 (881 GB, 1+0): Rebuilding 40% complete).

Question now. Is it safe ot replace the failed drive at this stage or should I wait until logical drive will be rebuild with online spare?

Seems to me it is safe to replace disk during rebuild but have no way to check it out and don't wan't to risk data loose in case I'm wrong.

Thanks in advance for reply!

[Stephen Kebbell]

No! Wait until the rebuild is finished!

Regards,
Stephen

[Luk Vandenbussche]

Wait till the rebuild is completed before replacing the disk.

[Adam Olejniczak]

Thanks for replies. I do wait everytime. Was just trying to understand why it is needed.

As I see it if disk fails it provide no data, On-Line spare has no data yet. So data on online spare is being rebuild based on data present on other disks in the array.

So situation is like this that nor the failed disk nor the Spare has data yet. Shouldn't I be able to replace failed drive then? After all it is "not working" anyway. Shouldn't controller just handle the rebuild process in proper order even if I will replace the drive during the time spare is rebuilding?

Why I ask is that if the our statement from posts above would be true then one need access to diagnose tools and/or OS to check what is the rebuild status at current and couldn't rely just on optical investigation of disk lights. It just sounds bit not right to me.

[Steven Clementi]

Adam:


While the controller has a Rebuild in Progress, it is always better to wait to swap the failed disk especially since if it is already rebuilding on the spare, the swap will do you no good. It will continue to rebuild on the spare.


You can check the progress of the rebuild using the Array Configuration Utility. You can also get some Diagnostic feedback from the Array Diagnostic Utility. (ACU/ADU)

Is it safe to replace the failed drive? Yea, it probably is... but it is safer to let your redundancy rebuild before forcing your controller to do something new.


Steven

[Stephen Kebbell]

Hi,

while a rebuild is in progress, the middle LED of the disk being rebuilt blinks once per second. All other disks (apart from the failed one) have this LED on (not flashing). When that blinking stops, the rebuild should be finished. Also, if you use Systems Insight Manager you will normally get an event from the server when the rebuild is finished. The Insight Agents can also provide this info, without the notifications.

Regards,
Stephen