Intel matrix RAID keeps rebuilding mirror array

[Chris_B]

I've got an Asus P5E mobo, and I'm using the on-board Intel Matrix RAID ICH9R chipset and drivers under Vista x64 to manage up two RAID arrays, one non-RAID drive and a SATA BD drive.

The OS array is 2x WD1500ADFD drives in a RAID 1 mirror pair. This seems ot be fine.

The DATA array is 2x WD7500AAKS drives in a RAID 1 mirror pair. This is the one that's causing me trouble. Twice now, for no discernable reason, this array hase gone into rebuild, which takes a few hours. I daren't shut down or reboot during the rebuild, so I just have to sit an wait it out, which is annoying when I'm trying to install drivers, patches and apps that all want a reboot.

Any ideas why it would be doing this? I can't find any error logs, and there's nothing to indicate a drive problem in the Windows event logs.

Is there a log in the Matrix Array that I can turn on to find out what's causing this?

The installed kit is 7.5.0.1017, and the driver and option ROM versions match this.

[ QUOTE ]

System Information

Kit Installed: 7.5.0.1017

Kit Install History: 7.5.0.1017

Shell Version: 7.5.0.1017

OS Name: Microsoft® Windows Vista™ Business

OS Version: 6.0.6001 Service Pack 1 Build 6001

System Name: ASUSP5E

System Manufacturer: ASUSTeK Computer INC.

System Model: P5E

Processor: Intel® Core2 Extreme CPU X9650 @ 3.00GHz

BIOS Version/Date: American Megatrends Inc. 0601 , 01/24/2008

Language: ENU

Intel® RAID Technology

Intel RAID Controller: Intel® ICH8R/ICH9R SATA RAID Controller

Number of Serial ATA ports: 6

RAID Option ROM Version: 7.5.0.1017

Driver Version: 7.5.0.1017

RAID Plug-In Version: 7.5.0.1017

Language Resource Version of the RAID Plug-In: 7.5.0.1017

Create Volume Wizard Version: 7.5.0.1017

Language Resource Version of the Create Volume Wizard: 7.5.0.1017

Create Volume from Existing Hard Drive Wizard Version: 7.5.0.1017

Language Resource Version of the Create Volume from Existing Hard Drive Wizard: 7.5.0.1017

Modify Volume Wizard Version: 7.5.0.1017

Language Resource Version of the Modify Volume Wizard: 7.5.0.1017

Delete Volume Wizard Version: 7.5.0.1017

Language Resource Version of the Delete Volume Wizard: 7.5.0.1017

ISDI Library Version: 7.5.0.1017

Event Monitor User Notification Tool Version: 7.5.0.1017

Language Resource Version of the Event Monitor User Notification Tool: 7.5.0.1017

Event Monitor Version: 7.5.0.1017

Array_0000

Status: No active migration(s)

Hard Drive Write Cache Enabled: Yes

Size: 279.4 GB

Free Space: 0 GB

Number of Hard Drives: 2

Hard Drive Member 1: WDC WD1500ADFD-00NLR5

Hard Drive Member 2: WDC WD1500ADFD-00NLR5

Number of Volumes: 1

Volume Member 1: System

Array_0001

Status: Rebuilding

Hard Drive Write Cache Enabled: Yes

Size: 1397.2 GB

Free Space: 0 GB

Number of Hard Drives: 2

Hard Drive Member 1: WDC WD7500AAKS-00RBA0

Hard Drive Member 2: WDC WD7500AAKS-00RBA0

Number of Volumes: 1

Volume Member 1: Data

System

Status: Normal

System Volume: No

Volume Write-Back Cache Enabled: No

RAID Level: RAID 1 (mirroring)

Size: 139.7 GB

Physical Sector Size: 512 Bytes

Logical Sector Size: 512 Bytes

Number of Hard Drives: 2

Hard Drive Member 1: WDC WD1500ADFD-00NLR5

Hard Drive Member 2: WDC WD1500ADFD-00NLR5

Parent Array: Array_0000

Data

Status: Rebuilding: 57% complete

System Volume: No

Volume Write-Back Cache Enabled: No

RAID Level: RAID 1 (mirroring)

Size: 698.6 GB

Physical Sector Size: 512 Bytes

Logical Sector Size: 512 Bytes

Number of Hard Drives: 2

Hard Drive Member 1: WDC WD7500AAKS-00RBA0

Hard Drive Member 2: WDC WD7500AAKS-00RBA0

Parent Array: Array_0001

Hard Drive 0

Usage: Array member

Status: Normal

Device Port: 0

Device Port Location: Internal

Current Serial ATA Transfer Mode: Generation 1

Model: WDC WD1500ADFD-00NLR5

Serial Number: WD-WMAP41863073

Firmware: 21.07QR5

Native Command Queuing Support: Yes

Hard Drive Write Cache Enabled: Yes

Size: 139.7 GB

Physical Sector Size: 512 Bytes

Logical Sector Size: 512 Bytes

Number of Volumes: 1

Volume Member 1: System

Parent Array: Array_0000

Hard Drive 1

Usage: Array member

Status: Normal

Device Port: 1

Device Port Location: Internal

Current Serial ATA Transfer Mode: Generation 1

Model: WDC WD1500ADFD-00NLR5

Serial Number: WD-WMAP41866229

Firmware: 21.07QR5

Native Command Queuing Support: Yes

Hard Drive Write Cache Enabled: Yes

Size: 139.7 GB

Physical Sector Size: 512 Bytes

Logical Sector Size: 512 Bytes

Number of Volumes: 1

Volume Member 1: System

Parent Array: Array_0000

Hard Drive 2

Usage: Non-RAID hard drive

Status: Normal

Device Port: 2

Device Port Location: Internal

Current Serial ATA Transfer Mode: Generation 1

Model: WDC WD360ADFD-00NLR5

Serial Number: WD-WMANT1118700

Firmware: 21.07QR5

Native Command Queuing Support: Yes

System Hard Drive: No

Size: 34.4 GB

Physical Sector Size: 512 Bytes

Logical Sector Size: 512 Bytes

Hard Drive 3

Usage: Array member

Status: Normal

Device Port: 3

Device Port Location: Internal

Current Serial ATA Transfer Mode: Generation 2

Model: WDC WD7500AAKS-00RBA0

Serial Number: WD-WCAPT0634385

Firmware: 30.04G30

Native Command Queuing Support: Yes

Hard Drive Write Cache Enabled: Yes

Size: 698.6 GB

Physical Sector Size: 512 Bytes

Logical Sector Size: 512 Bytes

Number of Volumes: 1

Volume Member 1: Data

Parent Array: Array_0001

Hard Drive 4

Usage: Array member

Status: Normal

Device Port: 4

Device Port Location: Internal

Current Serial ATA Transfer Mode: Generation 2

Model: WDC WD7500AAKS-00RBA0

Serial Number: WD-WCAPT0628796

Firmware: 30.04G30

Native Command Queuing Support: Yes

Hard Drive Write Cache Enabled: Yes

Size: 698.6 GB

Physical Sector Size: 512 Bytes

Logical Sector Size: 512 Bytes

Number of Volumes: 1

Volume Member 1: Data

Parent Array: Array_0001

CD/DVD Drive 0

Device Port: 5

Device Port Location: Internal

Current Serial ATA Transfer Mode: Generation 1

Model: HL-DT-ST BD-RE GGW-H20L

Serial Number: K1681PI1227

Firmware: YL02

[/ QUOTE ]

[CarMad]

Not seen that with mine, just look and I think these are the latest drivers...

7.8.0.1012 11/9/2007

Worth an update and see what happens ?

[Chris_B]

Just tried updating to latest Intel X38 chipset INF files, and got a BSOD and reboot for my troubles. The System RAID 1 volume is now verifying and repairing, and despite being 150GB mirrored, seems slower to repair than the 750GB Data volume does.

Shows how much of a performance improvement RAID 1 normally makes though - this thing crawls with heavy disk IO on the array.

Dare I update and reboot during a rebuild?

[craigyb]

These are just mirrors? why not break the mirrors, get the system stable, update all your drivers etc and add the second disk to each mirror.

[bullett]

RAID 1 is for security not performance. If anything RAID 1 will perform slower than non RAID and certainly slower than RAID 0.

I had a similar problem and it turned out to be a fault disk. Replaced and all fine now.

[Chris_B]

[ QUOTE ]

These are just mirrors? why not break the mirrors, get the system stable, update all your drivers etc and add the second disk to each mirror.

[/ QUOTE ]

Might well do that. Updated drivers last night, using latest Intel (not Asus) Matrix RAID driver set, and so far no issues.

The Option ROM BIOS still shows as the old version though, and the main Asus BIOS update doesn't update it, so there must be a separate update for the RAID BIOS somewhere, but I'm struggling to find it.

[Chris_B]

Another rebuild this morning.

Also, Problem Reports & Solutions (which seems to have more of the former than the latter!) also says there was a "Video hardware error":

[ QUOTE ]

Product

Windows

Problem

Video hardware error

Date

27/03/2008 08:03

Status

Solution Available

Description

A problem with your video hardware caused Windows to stop working correctly.

Problem signature

Problem Event Name: LiveKernelEvent

OS Version: 6.0.6001.2.1.0.256.6

Locale ID: 2057

Files that help describe the problem (some files may no longer be available)

WD-20080327-0802.dmp

sysdata.xml

Version.txt

Extra information about the problem

BCCode: 117

BCP1: FFFFFA8008F8A010

BCP2: FFFFFA60028165F0

BCP3: 0000000000000000

BCP4: 0000000000000000

OS Version: 6_0_6001

Service Pack: 1_0

Product: 256_1

Server information: afeef0da-9038-4a17-bc05-e03f7a10f3d6

[/ QUOTE ]

I keep thinking perhaps I'll just go back to the old PC; it was slower, but it worked a lot more often...

[Chris_B]

Another incident yesterday. I was typinga forum post and everything started going r e a l l y s l o w l y . . .

I mrean really slow - sort of type something and it appears 30 seconds later slow.

Then the screen blinked a bit, Vista burped and the Matrix RAID started a rebuild, and I got a ballon in the system tray that said the video driver had experienced a problem and recovered, then about 30 buffered keystrokes and mouse clicks played out.

I've been having the odd problem where the Reclusa (or is it Habu?) illuminated keyboard just doesn't work at boot. I have to unplug it and plug it back in again to get it to work. I also had one or two issues with the Razer mouse (had to move it to a different USB port to make it reliable), and an external USB2 drive (powered though) has been a bit erratic during large file copies.

Think I've got a pit of a power issue somewhere? I guess I'll be fetching the PSU tester from the office and giving the Coolermaster a run over. It's 830w if I remember rightly, so I'm hoping it's sufficient to juice up a quad core Extreme, 8GB of RAM, an 8800GTX and five SATA HDDs (three 10k).

Are these the kind of symptoms a power rail running low on juice would give? Which is the most likely culprit - 12v? This is a modular PSU, so should I plug more cables in and split the load between more outputs on the PSU?

I wonder if I'm going to have to go with a PSU with split 12v rails.

[Thorburn]

Try Memtest, I used to have all kinds of NVIDIA driver reset issues that turned out to be down to a faulty DIMM.

[Chris_B]

Cheers, will give that a go.

[Chav]

onboard raid is often dodgy. I'd avoid with a bargepole.

def test the psu.

multiple psus may be in order if you've got ninja graphics setup? simple, effective but need more case space obviously.

just noticed your hardware spec.... 5 sata h/disks no problem. so sounds like a dodgy psu. try running the graphics from a separate psu to everything else.

is the system sufficiently cooled?

5 hard disks need a big case. to keep air cool.... open the case or put the drives to the side outside etc and test?

[Chris_B]

Well, I ran memtestx86+ v2.01 on it last night, and got through a single run with no errors, which took longer than it's generally been running before hiccuping.

Next stop, PSU tester. After that, who knows...

One thought regarding power - I have a Razer keyboard and mouse, both LED illuminated. That seems like a lot of power to draw from a USB hub on the mobo - think this could be related? I also have a Wacom A4 pad, a USB printer, a scanner and a UPS hooked up, as well as en external USB2 HDD.

I have noticed the illumination on the keyboard and mouse blinks on and off at POST and during Windows startup, but sometimes the keyboard flickers wildly as well. Think I'm just drawing too much juice through the USB ports, and the mobo is struggling to cope?

[Thorburn]

Not sure, the USB is controlled by the motherboard ICH, as is the RAID controller, you could try a powered USB hub.

[Chris_B]

Windows doesn't think anything's drawing too much power BTW, the mobo hubs all show as up to 500mA per port, and the most I'm drawing on any device seems to be the Razer at 300mA.

Still, makes sense to give it a try, doesn't it? It's not like a USB2 powered hub is a lot of money or anything (still, it will be annoying if the mobo can't support power to the 12 or so USB ports it has...)

[Chris_B]

Well, I unplugged my non-essential USB devices to see if it made a difference. It did - while unplugging, nVidia System Guard noted that due to a power shortage, the performance of the 8800 was reduced automatically - so it does seem to be power problems!

I bought a cheap and nasty Advent powered USB2 hub from PC World, and after having to return it and replace it because the PSU was faulty, I ended up with a working 7-port powered hub.

I've now got the Razer mouse and keyboard on different hubs on the mainboard, and the 7-port hub on another, and into that powered hub I've plugged the Wacom, WD external drive, Pixma printer, Canoscan scanner, APC UPS and I have two ports spare. I still have two ports on the outside of the case (plugged into mobo headers) and I've had a USB flash drive and an iPod in there recently, and no problems.

I haven't had a RAID rebuild since unplugging everything from the mobo, so fingers crossed that was it.

So, is it the mobo power regulator, or the PSU? The total watts should be OK on the PSU (830 IIRC), but perhaps a specific rail or two are low?

If it's the mobo, Asus need a stern finger wagging - what's the point of having 12 USB ports (rear plate and headers) that all report 500mA max when you can't use them all concurrently without tripping up the RAID or PCIe?

[Chav]

hiya,

i'd use multiple psu's. much more reliable than 1 big one (whether modular or not)

and should always use powered hubs.

[Chris_B]

Yeah, I did think about one of those Antec cases with two PSUs in it, for 1200w of power, but thought it was overkill for a Core2 PC with a single GPU - this stuff's supposed to be getting more poswer efficient!

Anyway, I ended up with the Coolermaster which came with an 830w PSU. Now I got to see what might fit in it, seeing as I've nicked my spare Tagan for another self-build that had a flaky PSU.

I don't see the point in splashing out on a PSU though, if it turns out it's a mobo fault that too much drain on the USB ports causes instability.

[Thorburn]

Before buying anything try swapping around the connectors you are using for the devices, I expect the supply has multiple 12V rails so you may simply have too many devices attached to one rail. An 830W should be ample as I run a similar setup (Quad-core, 8GB DDR2-800, 8800GTX, etc) off a 520W supply, draw at the wall reaches about 300W max.

[Chris_B]

It's a modular PSU and IIRC, I have a single PCIe 6-pin connector for the 8800, two for SATA drives (three 10k and two 7.2k) between them, and one for the BD-ROM drive, plus the ATX (24-pin) and 12V (4-pin I think) connectors.

That's it.

I guess I could try swapping some of the modular plugs around on the outlets on the PSU, as I've no idea (and it's not documented) what outlet shares what 12v rail.

[Chris_B]

Ar5eholes!

Booted it this evening, and it's clicked and whirred and chuntered, the screen flickered and I could hear the hard drives spin down and up again.

Logged in, and Intel Matrix RAID says a drive has been removed, and the BD burner access light starts flashing away, and the PC runs very slowly and jerkily (mouse movements etc).

30 seconds later, back to normal speed Matrix RAID says drive inserted, and a rebuild starts, and speed returns to normal.

I can still hear one of the drives powering down and up again every now and then.

Sounds like a new PSU might be on order soon. What's a good PSU with good 12v power rail loading that works with a Coolermaster CM690 case?

Sigh.

[Chris_B]

This looks like the PSU that was bundled with the case: http://www.ebuyer.com/product/132289 or http://www.coolermaster.com/products/product.php?act=detail&id=35

Should be an OK one, shouldn't it? Six +12v rails at 18A (two at 28A - shared with PCIe?), ATX 12V v2.3 and EPS 12V v2.91 compliant, >85% efficiency and all that stuff. Doesn't seem like some cheap piece of junk. Perhaps it's just a duff one of a good design?

Because I can't be bothered to send the whole case back, or even to have the conversation with ebuyer, I might just buy one of these and send the duff PSU back in the box for a refund when it arrives. Bit naughty I know, but c'mon - I paid for a PSU that is duff and I'll send a duff one back, without the need to rip my PC to bits.

Are Coolermaster generally OK?

[Chris_B]

Well, I bought the CM 830W PSU, and fitted it at the weekend.

Booted first time, RAID started rebuilding but that's OK because the BIOS-level screen had it flagged as being in rebuild state anyway, and not a hiccup since. I used it for a few hours yesterday evening, and managed to install Office from DVD, download hundred of MB of Office updates, and didn't get any more odd effects.

Hopefully, that's it.