Hard Drives Keep Dropping Out, System Reboots?

Discussion in 'General Hardware' started by madmatt, May 13, 2007.

  1. madmatt

    madmatt Bow Down to the King Political User

    Messages:
    13,312
    Location:
    New York
    I have a weird problem that I have never encountered before. At first I thought it was a bad hard drive but it has only gotten weirder.

    This is my Exchange Server running on Windows Server 2003 R2 x64. I have two 500GB WD RE2 drives that were configured to mirror (using the Disk Management snap-in). This has been working since January flawlessly.

    Last Monday I look over at the server and see the hard drive light pinged. I login to find that the mirror had failed. Assuming the drive has failed I turn off the server, check the connections, and turn the server back on. It boots up perfectly. I recreate the mirrors.

    A few days later the same thing happens. This time I reboot the server and it gets stuck at POST telling me no drive can be found at SATA Port 0.

    I turn off the server and remove the drive at port 0, replacing it with the drive loaded on port 1. Now I am convinced the drive is bad.

    I power on the server, make the changes in the BIOS, and everything fires up just fine. Now here's where it gets weird.

    Last night I am relaxing on the couch watching COPS (not that you cared) and I look over at my notebook and it says connection lost and that it is trying to reconnect to the Exchange Server. I walk into my office, turn on the display, and find my server sitting at POST saying SATA drive not found at Port 0. WTF?

    I turn off the server and let it sit for a few moments. I power it back on and it loads up fine. I check the Event Viewer and nothing, no errors or warnings.

    Any ideas?
     
  2. X-Istence

    X-Istence * Political User

    Messages:
    6,498
    Location:
    USA
    Download SMART utility to find out if there are any premature warnings about failures for the drive.

    Ive had the same thing happen, and then the next time I powered the drive down, and back up it died.
     
    madmatt likes this.
  3. madmatt

    madmatt Bow Down to the King Political User

    Messages:
    13,312
    Location:
    New York
    I've run the WD Data Lifeguard Diagnostics on both drives. Including SMART. It found nothing wrong and said both drives passed.

    The server has been rebooted several times during these on going issues.
     
  4. Steevo

    Steevo Spammer representing. Political User Folding Team

    Messages:
    2,566
    Onboard drive controller or add in card?

    If you have any additional ports you might try them.
     
  5. madmatt

    madmatt Bow Down to the King Political User

    Messages:
    13,312
    Location:
    New York
    On board. I will be purchasing an add-in controller after this mess.

    The server was purchased in December and the drives were purchased in January.
     
  6. Steevo

    Steevo Spammer representing. Political User Folding Team

    Messages:
    2,566
    Here recently the server I built a few years ago has been eating parts like candy. Mostly due to ****ty power in our building.


    four 300GB HDD's
    One power supply (It actually threw sparks out!!).
    Onboard ICHR7 controller went with the first set of drives.



    But $ for performance I like the Rocket Raid controller we got, minus the part where my PCIe slot was taken. RAID 5 and it really does a good job on rebuilds and is easy to manage.
     
    madmatt likes this.
  7. LordOfLA

    LordOfLA Godlike!

    Messages:
    7,027
    Location:
    Maidenhead, Berkshire, UK
    The onboard SATA controller is likely on the way to the grave. Get yourself a 3ware 8000 2 port (9000 series if you want to be flash) to run the drives and provide for hardware raid should you want to go that way in the future.

    There are cheaper controller cards available however we've found here at work that cheaper cards are none too helpful when a disk in an array dies. Our highpoint rocket raid insisted on clearing a raid5 array of data when we wanted to rebuild it after replacing a disk. go figure.
     
    madmatt likes this.
  8. madmatt

    madmatt Bow Down to the King Political User

    Messages:
    13,312
    Location:
    New York
    Thanks Lord. As I mentioned I might have a need for more information after making a decision.

    It appears I will need a controller card because I, too, believe it is a problem with the onboard controller. Will I have to rebuild the system or is there an easy way to migrate to this controller card (the 8000 or 9000 series card) without losing everything?

    Which one do you recommend? The 8000 or 9000?

    EDIT: and WHICH model would you recommend. There are several different ones available, some that are very expensive.
     
    Last edited: May 14, 2007
  9. LordOfLA

    LordOfLA Godlike!

    Messages:
    7,027
    Location:
    Maidenhead, Berkshire, UK
    The model to get depends on what you want.

    The 8000 series is nice and cheap for 2/4/8 ports. Has varying amounts of on-controller cache and supports battery backup.

    The 9000 series supports controller chaining for massive arrays if you have the space for them and the associated disks. Imagine 4x 24 disk controllers chained :) Obviously this series costs more.

    For a simple 2 disk exchange array though I'd say go with the 8006-2LP.

    Edit: Forgot to add, we successfully took a live array from a highpoint card to a 3ware 9000 series by making sure we had the disks wired up in the exact same order we took them off the highpoint. Added the new disk and away it went.
     
    Last edited: May 14, 2007
  10. madmatt

    madmatt Bow Down to the King Political User

    Messages:
    13,312
    Location:
    New York
    So Lord, you're saying I *should* be able to install the controller and install my current drive on PORT 0 and make it through without having to start from scratch?
     
  11. X-Istence

    X-Istence * Political User

    Messages:
    6,498
    Location:
    USA
    That is what I read.
     
  12. Geffy

    Geffy Moderator Folding Team

    Messages:
    7,805
    Location:
    United Kingdom
    *should* maybe more like *could*. If you can make a backup of your data beforehand. Switching from one RAID card to another can sometimes work and sometimes not. Better to be safe than buggered ;)
     
  13. madmatt

    madmatt Bow Down to the King Political User

    Messages:
    13,312
    Location:
    New York
    I'd be going from the onboard SATA port with no RAID to a RAID controller.

    Dell is telling me I am better off starting from scratch and restoring a backup after it is in place.
     
  14. LordOfLA

    LordOfLA Godlike!

    Messages:
    7,027
    Location:
    Maidenhead, Berkshire, UK
    Well how vital is your data?

    We had a requirement to keep the array live as it had died during its first backup run. We just cabled up the drives exactly and away it went.

    If you can take a backup do so. Windows should still operate the mirror - its smart. THe card will give you the option of a hardware setup down the line. Which is the better option overall.
     
  15. Steevo

    Steevo Spammer representing. Political User Folding Team

    Messages:
    2,566
    I would have liked to go hardware support, but with the extra cost, and the fact that on a 3.8Ghz server 2% makes no tangable difference......



    But I completely agree that if you have the funds and need, do it.
     
  16. madmatt

    madmatt Bow Down to the King Political User

    Messages:
    13,312
    Location:
    New York
    The server is my Domain Controller and my Exchange. All of my ISO's are also stored on the server. So the data is rather important. However, I operate in Cached Exchange Mode so my mailbox can be synced easily. And I can copy the rest off to my external hard drive. It's just that I am a perfectionist and I finally got the server (the OS and Exchange that is) running the way I want it running.

    However, I know I need to spend the money (I also purchased another 4GB of RAM) so I will buy a the hardware (a PERC 5i has been recommended to me for the SC1430). After that and I save up more I will be buying a backup drive.
     
  17. LordOfLA

    LordOfLA Godlike!

    Messages:
    7,027
    Location:
    Maidenhead, Berkshire, UK
    I'm not sure you want the PERC over the 3ware. Compare the two and see how you feel. The advantage to the PERC is that windows ships with the drivers.