1 Reply Latest reply on Nov 9, 2011 3:40 AM by Flying_Kiwi

    BIOS often declares RAID-0 array "failure" prematurely when booting?

    Huzer

      Hello!  Thanks to all for taking the time to read this!

       

      Roughly 40% of the time as I boot this particular machine, the onboard Intel RAID BIOS far too quickly declares that my 2-member RAID-0 boot array has "failed" (and would thus be unbootable), even though if it had waited just a few more seconds eveything would have been fine!  Whether I perform a cold or warm boot doesn't matter at all.  The other 60% of the time everything's perfect.

       

      My mobo is an Intel DQ35MP with the latest system BIOS (JOQ3510J.86A.1143.2010.1209.0048), and the BIOS Drive Configuration is set to "Native / RAID / Recovery=Disabled". I'm running 32-bit Windows XP Pro/SP3, and my system resides on a RAID 0 array comprised of two Western Digital Caviar Black 500GB drives (model WD5002AALX), each of which passes all the diagnostics and benchmarks I've tried.

       

      Here's the irritating 40% scenario:

       

      1. I either press the power button (for a cold boot) or have finished a previous Windows session with a "Shutdown->Restart" menu (warm boot).
      2. The main Intel BIOS ID is displayed briefly, then the Intel RAID BIOS screen comes up.
      3. The RAID 0 boot array is marked with the word "Failure" in red, and only one of the two RAID drives shows up and is marked as a "non-member".
      4. The prompt to enter RAID setup comes up, and I enter Control-I to enter the setup screen.
      5. Approximately half the time, by the time the setup screen shows up, the array is listed as "Normal" (in green) already!  The other half of the time, I have to twiddle away a few more seconds by selecting pointless menu options and then pressing the escape key to return to the main menu again, by which time the array is reported to be "Normal" and ready to go (the amount of time I need to waste varies from 1 up to 10 seconds).  NOTE that no actual operations are performed at all -- all that matters is that I delay the boot process by a few more seconds to allow both disks to become ready and be recognized as such by the RAID BIOS.
      6. I then exit the RAID BIOS and the boot continues succesfully every single time.

       

      One main complaint is that since this occurs randomly (absolutely no pattern has emerged), I have to babysit this machine every time it boots, which is unacceptable to me.

       

      I would have thought that increasing the disk spinup delay in the system BIOS would solve this problem, since all that is needed is a few more seconds delay just prior to the RAID BIOS firmware startup, but I've learned that the spinup delay occurs far too late in the process (and is intended for IDE drives rather than SATA drives anyway), so that's useless.

       

      I've also tried alternating the Drive Configuration settings in the sytem BIOS between "Native / RAID / Rapid Recover Technology=Enabled" (which runs RAID 0/1 firmware) versus "Native / RAID / Rapid Recover Technology=Disabled" (which runs the "any RAID including RAID5" firmware), but there's no apparent difference as far as the 40% failure scenario goes.

       

      Finally, here are my questions:


      What's going wrong here?


      How can I delay or slow down the RAID BIOS firmware in order to keep it from testing the array drives too quickly and thus jumping to the wrong (failure) conclusion?


      These are new, high-quality 7200-rpm Caviar Black drives, shouldn't they both spin-up and be ready fast enough for the RAID BIOS?  How can I determine whether they spin up fast enough according to some objective measure such that I can turn to Western Digitial for repair/replace?


      Perhaps the RAID firmware versions are too old (as Sandra and other system evaluation tools keep insisting), but Intel makes it impossible to find and update just the RAID firmware alone, and I already have the latest (December 2010) sytem BIOS.  Is this an approach worth following, or is it a dead end?

       

       

      More info: I'm running Intel's Rapid Storage Technology software, currently version 10.6.0.1002.  Here's the relevant portion of the System Report it generates:

       

      System Information

      OS name:  Microsoft Windows XP Professional

      OS version:  5.1.2600 Service Pack 3 2600

      System manufacturer:  INTEL_

      System model:  DQ3510J_

      Processor:  GenuineIntel x86 Family 6 Model 23 Stepping 6 2.989  GHz

      BIOS:  Intel Corp., JOQ3510J.86A.1143.2010.1209.0048

       

      Intel® Rapid Storage Technology

      Kit installed:  10.6.0.1002

      User interface version:  10.6.0.1002

      Language:  English (United States)

      Intel RAID controller:  Intel(R) Desktop/Workstation/Server Express Chipset SATA RAID Controller

      Number of SATA ports:  6

      RAID option ROM version:  8.9.0.1023

      Driver version:  10.6.0.1002

      ISDI version:  10.6.0.1002

       

      Device Information

      Name:  Array_0000

      Size:  932 GB

      Available space:  0 GB

      Disk data cache:  Enabled

      Number of volumes:  1

      Volume member:  RAID_CAVBLK

      Number of disks:  2

      Array disk:  WD-WCAYUX735573

      Array disk:  WD-WCAYUX820887

       

      Name:  RAID_CAVBLK

      Status:  Normal

      Type:  RAID 0

      Size:  932 GB

      Data stripe size:  64 KB

      Write-back cache:  Enabled

      System volume:  Yes

      Initialized:  Yes

      Verification errors found:  0

      Blocks with media errors:  0

      Physical sector size:  512 Bytes

      Logical sector size:  512 Bytes

      Parent array:  Array_0000

      Number of disks:  2

      Array disk:  WD-WCAYUX735573

      Array disk:  WD-WCAYUX820887

       

      Disk on port 0

      Port location:  Internal

      Status:  Normal

      Type:  Hard disk

      Usage:  Array disk

      Size:  466 GB

      Serial number:  WD-WCAYUX735573

      Model:  WDC WD5002AALX-00J37A0

      Firmware:  15.01H15

      System disk:  No

      Disk data cache:  Enabled

      Native command queuing:  Yes

      SATA transfer rate:  3 Gb/s

      Physical sector size:  512 Bytes

      Logical sector size:  512 Bytes

       

      Disk on port 1

      Port location:  Internal

      Status:  Normal

      Type:  Hard disk

      Usage:  Array disk

      Size:  466 GB

      Serial number:  WD-WCAYUX820887

      Model:  WDC WD5002AALX-00J37A0

      Firmware:  15.01H15

      System disk:  No

      Disk data cache:  Enabled

      Native command queuing:  Yes

      SATA transfer rate:  3 Gb/s

      Physical sector size:  512 Bytes

      Logical sector size:  512 Bytes

       

       

      Thanks for your patience and attention, folks!

        • 1. Re: BIOS often declares RAID-0 array "failure" prematurely when booting?
          Flying_Kiwi

          My first thoughts were that maybe your HDDs had different firmware versions on THEM but thanks to your thoughtful inclusion of the Intel  RST report, I see that is not the case. When you do get this problem, is it always the same drive that fails or does that vary? If it's the same one every time, perhaps that particular disk has a fault? You are right that the RAID firmware for the motherboards controller comes as part of the BIOS so without Intel developers making a specific update for you (if one is even available for your hardware), you already have the latest and greatest for that

           

          It appears you've already tried the most obvious idea of manually putting in a HDD Pre-delay without success so it'd be worth going to WD about this as often HDD firware updates can resolve issues such as this. Be aware that drive manufacturers do not usually like giving such updates out willy nilly (or even advertising such availability on their websites). The last time I obtained such an update (from a different HDD manufacturer but for a RAID related issue), they wanted all my drive details including serial number and current firmware revision etc so make sure you have everything ready when you contact them. Perhaps they have newer firmware that they can provide you with which will sort the matter out - otherwise your HDD supplier may be some assitance.