I don't have an answer for you as my experience with Linux software raid is 
small.  Another source you can go to for answers is the linux-raid mailing 
list.  I subscribe to it and have seen these types of questions come 
through and the developers post answers.

Send an email to majordomo@vger.kernel.org with the command "subscribe 
linux-raid" (no quotes) in the body of the email.

Alan

At 02:28 PM 7/12/02 +0000, you wrote:
>Hi,
>
>This is my first post to this mailing list...I subscribed because of this
>issue I'm having and need help....
>
>First off, let me explain what I have (had) in place..
>
>An Intel 440GX server board with 2 processors and onboard LVD SCSI 
>(aic7xxx).
>A Promise UDMA controller (PDC20267 - don't remeber which model exactly, but
>that is the driver the kernel uses for it)
>7 IDE drives (Maxtor 4D080H4 x 3; Seagate ST380021A; Maxtor 98196H8; WDC
>WD800BB-32BSA0; WDC WD800AB-22BTA0)
>3 SCSI Drives (IBM and Seagate)
>
>I initially set up the system with a SuSE 7.3 install - 2.4.10 and have since
>compiled a custom kernel @ 2.4.18
>
>I have the 7 drives configured in Raid 5 with no spares for about 400 GB of
>storage (they are all 80 GB drives)
>
>There is an APC UPS (400VA capacity) connected to the system.
>
>The server was running fine until last night when it froze without response
>from pings or physically at the console...I hit the reset button to bring it
>back up...
>
>Now, the event counter for the raid 5 array lists 3 different things - 5
>drives show event counter to be 00 00 00 38, 1 drive shows 00 00 00 28 (hda),
>and 1 drive shows 00 00 00 37 (hdd).
>
>md says hdg1 is freshest and kicks "non-fresh hda1 from array"..it then says
>"kicking faulty hdd1" and "not enough operational devices for md0 (2/7
>failed)"
>
>Even if hdd is completely bad, I can't afford to lose 250 GB of data...if 
>hda1
>is out of sync, is there anyway to force it to accept it (with a few corrupt
>files maybe)...  I read that someone manually edited the event counter to
>allow it to think the drive is ok, I can't find the event counter when
>looking at the drive in hex (using Microscope Diagnostics to view drive)...I
>have no clue where on the drive to look.
>
>Someone said that using mkraid --force is the way to go because you can force
>a new superblock, that is contingent on having /etc/raidtab up to date...it
>seems to be empty for me...don't know why or where it went...I tried to
>recreate it manually with a typical raid 5 configuration, it keeps telling me
>invalid chunk size when I do an mkraid --force --configfile /etc/raidtab
>/dev/md0  (I put in 128k and it thinks I mean 128MBl; I put just 128 and it
>does the same...looked at man pages and followed their example, same
>problem)...
>
>Anyhow, the event counters are still the same as they were, so I don't think
>the mkraid actually wrote anything due to the errors.  I have 1 extra 80 GB
>drive that I can use to make a backup of an existing drive (I don't have 7
>though) -- it is a Maxtor tho, and the 2 failed drives are both Western
>Digitals....
>
>Any ideas or suggestions?  Am I barking up the wrong proverbial tree?  Should
>I shoot the server and tell my anime group (this was the anime server that
>they contributed funds to to build) of this catastrophe and risk a lynching?
>Any help is appreciated...thanks in advance.
>________________________________________________
>See http://PLUG.phoenix.az.us/navigator-mail.shtml if your mail doesn't 
>post to the list quickly and you use Netscape to write mail.
>
>PLUG-discuss mailing list  -  PLUG-discuss@lists.plug.phoenix.az.us
>http://lists.PLUG.phoenix.az.us/mailman/listinfo/plug-discuss

-
/------------------------------------------
|Alan Dayley             www.adtron.com
|Software Engineer       602-735-0300 x331
|ADayley@adtron.com
|
|Adtron Corporation
|3710 E. University Drive, Suite 5
|Phoenix, AZ  85034
\-------------------------------------------