RAID 10 problem - hard puzzle

[3d_compos]

RAID 10 problem - hard puzzle

Hello.

I have an issue with my RAID "10" setup (4 server class WD Caviar RE4 2 TB drives connected to Adaptec 2405 hardware controller).

The issue is that sometimes when the computer case is physically hit, the drive 0 in my RAID "10" array disconnects from the array. I can hear that this drive physically stops spinning. When I restart, the controller indicates that the array is crippled and rebuilds the array. (see a screenshot and log - pic 1 and 2). When finished, all is fine until the next occurance of the problem.

Actions done by me:
1. I changed the affected HDD to a new one - and yet the same was happening. So it was not a HDD issue.
2. I changed the power supply cable to another one - and yet the same was happening. So I guess it's not a power cable issue.

I suppose the two remaining possibilities are:
- a flaw in the special data cable MiniSAS SFF-8087 (see pic 3)
- a flaw in the controller itself.

What do you think, is it best first to try to replace the data cable and see if the problem is resolved, or should I just replace the whole controller (but it's more money)? Any other ideas for identifying the problem?

After I change the controller, will I then be able to keep the array, or will I be forced to set it up from scratch due to controller change?

PIC 1. Adaptec utility screenshot when rebuilding the array:

PIC 2: Adaptec log:

PIC 3: Data cable MiniSAS SFF-8087: http://www.superstorage.pl/product_info.php?products_id=87


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

CPU: Intel Core i7 2600K 3,40 GHz + (cooler) NOCTUA NH-D14
Mainboard: ASUS P8P67 EVO R3.0 Intel P67 LGA1155
RAID "10" controller: Adaptec 2405 KIT
HHDs comprising RAID "10" via Adaptec controller: 4x server class WD Caviar RE4 2 TB WD2003FYYS 64MB cache SATA/300
Power supply: BEQUIET Dark Power PRO P8
RAM: 4x Kingston HyperX DDR3 4GB 1600MHz XMP CL9 Genesis
Windows HDD: SSD OCZ Vertex 3 MAX SSD 120GB 550/500 MB/s 75k IOPs
Graphics card: GIGABYTE GeForce GTX 590 3072MB DDR5/384bit DVI/HDMI/DP PCI-E
OS: Windows 7 Professional BOX PL

 

[staticlag]

It's probably the cables.

But the real question is, why are you hitting your case? Can't you put it behind your desk?

 

[phis6]

Typically a lose cable connection. Either data or power cable I think.

 

[ND40oz]

I'd buy a new 8087 breakout cable to try next, you could use this type if you have molex plugs free on your power supply.

http://www.superstorage.pl/product_info.php?products_id=21

 

[TGK]

It's probably the cables.

But the real question is, why are you hitting your case? Can't you put it behind your desk?

If the drive spins down it sounds like power interruption to me. Unplugging the data interface with it on won't cause a spin down in my experience.

Please stop hitting a computer with drives spinning thousands of rpm. Just shouting loudly at drives has been shown to cause interruption to seeks.

 

[cookiesowns]

Doesn't sound like a cable issue to me. I had a backplane that did the same thing before, only on one drive though.

If I hot swapped a different drive, that drive will also powerdown. Ended up just replacing the chassis.

Derp.. OP didn't specify if he uses a backplane. If yes, then well give that a troubleshoot, if not, check the controler, cables if not already.

 

[patrickdk]

Backplanes should power up the drive in standby, and wait for the hba to init it.
Could have been an issue either way. But doesn't sound like the issue he is having.

Wonder if maybe the drive is resetting, and with everything else, it doesn't get enough power to turn on, when it's interrupted, that atleast would be easy to test.