wilddoktor

updated

14 years ago

W

wilddoktor

1 Rookie

•

24 Posts

0

770544

June 9th, 2012 21:00

Slowly blinking amber hard drive health light, poweredge 2900

Well, the support pages are down at the moment, otherwise I would have looked in the manual!

So, I've got the above mentioned pe2900 with a perc5/i and 4x146Gb SAS 15k drives in a RAID10 setup.  After installing Hyper-V Server 2008r2sp1, one of them now has the above mentioned slowly blinking amber "health" light...blinks green, then amber, then off, and repeats.

Am I losing a hard drive? 

Other info:

I deleted a RAID5 array, created the RAID10 array, and did a fast init. Then immediately rebooted to CD to install the OS. Not sure if I needed to do a full init first, but wanted to mention it.

Thanks for any help with this!

  • theflash1932

    11 Legend

    •

    16255 Posts

    15647

    1

    Posted June 9th, 2012 21:00

    Yes, best case scenario, you are losing a drive.  Amber/green indicates a "predicted failure".  This is most often a bad drive that has reported failing SMART information to the controller.  In some cases, this can signal a corrupt array - probable only if you did not delete and reconfigure a new array that had previously been in use before installing the OS.  Given the fact that you indicated having deleted the RAID 5 and created/initialized a new RAID 10, it is even more likely a failing drive.  You can run Online Diagnostics to confirm.

    You can access the manual directly here:

    support.dell.com/.../hom.pdf

  • theflash1932

    11 Legend

    •

    16255 Posts

    15647

    1

    Posted June 9th, 2012 22:00

    The "utility partition" is the partition where the diagnostics is installed to; it is installed on factory installs and can be installed/reinstalled by using the installation utility SBUU/SMTD.  It isn't necessary to have, and I usually don't bother when reinstalling a server.  Here is a link to the latest version of the bootable diagnostics (called 32-bit Diagnostics):

    http://www.dell.com/support/drivers/us/en/RC1232625/DriverDetails?DriverId=Y6D93&FileId=2731107576&DriverName=Dell%2032%20Bit%20Diagnostics%2C%20v.5118A0%2C%205118.3&productCode=poweredge-2900&urlProductCode=False (btw ... the Support site is pulling up everything fine for me)

    SBE Log Disable:  Every time the system corrects a single-bit error (SBE), as ECC memory is designed to do, it keeps track of how many.  If it logs some predetermined number between boots, then it stops logging them and issues the SBE Log Disable message to alert you to a potential issue with the memory.  Since you just installed it yesterday, it could be bad memory or could simply have a bad connection.  Before doing anything, I would reseat that DIMM, making sure there is nothing (dust, hair, etc.) in the slot, then run diagnostics on it.  If it fails, then it likely needs replacing.  If it passes, then it may be resolved.

    Another possibility is a memory bug in the BIOS; 9G servers (2900, 1950, etc.) were especially prone to memory errors with early versions of BIOS.  I would suggest making sure your system firmware (BIOS, ESM, RAID, HDD, NIC, DRAC) is up to date to prevent any number of other preventable issues from occurring.

     

  • wilddoktor

    1 Rookie

    •

    24 Posts

    15647

    0

    Posted June 9th, 2012 22:00

    Thanks a ton for the link...looking thru it now!

    I also just now also started getting a "E2111 SBE Log Disable DIMM 2" error on the lcd; (I added 32gb of ram yesterday). The manual says to Run System Diagnostics by pressing f10 during boot. Problem is, I've wiped the array...so there's no "system" partition.

    Any chance you have a link to a diagnostic I can run from a bootable cd / dvd?

  • wilddoktor

    1 Rookie

    •

    24 Posts

    15647

    0

    Posted June 10th, 2012 15:00

    Thanks so much flash! I downloaded the diags to a usb stick and booted from it...and the drive, indeed, has failed smart tests and will need replacing. Drat...it's out of warranty.

    The memory issue is strange. Upon testing, I get a big red warning saying "ecc critical threshold exceeded error dimm2." Then it says "The event log contains memory related entries that indicate degraded or disabled ecc functionality. Memory testing cannot continue until the indicated problems are corrected, the event log cleared and the system rebooted."

    Just for grins, I swapped dimm1 and dimm2 (they were bought as a matched pair); figured if the dimm is actually bad the error will follow it to dimm1. Nope; after swapping and re-running mem diags, I get the exact same error, about dimm2...even though dimm2 is now dimm1.

    Would this indicate a bad slot? 

    How do I clear the event log...what event log is it talking about?

  • wilddoktor

    1 Rookie

    •

    24 Posts

    15647

    0

    Posted June 10th, 2012 15:00

    Oops, my bad...now it says dimm1; guess I needed to wait a bit to run the test again.

    So...bad hdd, bad dimm. Now I know. :-)

    Thanks again so much, "flash"...I'm going to mark your post as the answers.

  • theflash1932

    11 Legend

    •

    16255 Posts

    15647

    0

    Posted June 10th, 2012 17:00

    Make sure to clear the ESM/Hardware Log before running the diagnostics; if there is a memory error in the logs, the diags will automatically fail.  The hardware log (also referred to as the SEL - System Event Log) it is talking about is the log of system events kept by the ESM/BMC - the onboard management chip.  You can clear the logs in CTRL-E or in OMSA (OpenManage Server Administrator) in Windows (System, Logs tab, Hardware sub-tab).  Then you can work to isolate the source of the fault (after also reseating memory DIMM's), whether it follows the DIMM or the slot.