Ive been using ver 4 beta 10 successfully for a while now...

 

Friday my drives reported that 2 drives were missing after I rebooted.  I unassigned one of the drives because I know there is no data on it and now when I reboot the server stays stuck on Mounting for all of the drives and 4 of the drives have the read value set to 24.  I have unplugged and plugged the server back in with the same results...  any help would be greatly appreciated.... 

Wow!  That is one sad syslog!

 

Disk1 and Disk2 are failing immediately to mount:  "disk1 mount error: 32" and "disk2 mount error: 32".  I have no clue as to why.

 

The 3 SATA drives (sda=Parity, sdc=Disk3, sdd=Disk4) are all showing multiple Exception EMASK errors.  They could be bad or failing drives, but that seems too coincidental.  It could possibly be a configuration problem, in the BIOS look for choices like IDE, Enhanced, AHCI, etc.  You should report your hardware, and perhaps someone with the same or similar motherboard can advise.  I'm wondering if there are any compatibility issues between your SIS chipset and your Promise cards.  Maybe not, and someone else with a similar board and cards can confirm successful operation.  You may have a failing motherboard or card.

 

Disk7 was told to mount, but does not appear to respond at all.

 

Disk8 is also told to mount, even though it is not assigned.  I assume it used to be assigned to hde, a WD250.  This is probably the reason for the hang, because it tries hard to mount it, gets some response, and reiserfs then reports problems with it.

 

Only Disk5 and Disk6 seem to be working correctly.  I would un-assign the parity drive, since there is no point until you have a stable system.  Re-assigning Disk8 to hde may get you past the mounting hang, but wouldn't solve the problem.  The data on the drives is probably fine, so a fresh start, a reset of your unRAID system seems the best advice I can give.  Others may have better advice as to your next steps.

 

Just curious, and probably unimportant, do you have a preference for 4.0 beta10 over 4.0 final?

 

No preference... just worked fine so I never changed it to the final version

 

a fresh start, a reset of your unRAID system

 

How do I accomplish this? Tom.. any clues?

Backup your flash drive first.  Then a 'fresh start' or reset would be accomplished by deleting the super.dat file and editing disk.cfg, both in the flash's config directory.  When editing disk.cfg, you will remove the diskN lines, leaving only the top title line and the spindown line.  Then boot unRAID and assign the desired drives.  By restarting this way, all other parameters should be unchanged.

 

I noticed in the release notes for 4.0 final that it lists the following as a change from beta10 to the final: 

Improvement: added Promise Ultra100 IDE support

 

There's a slight possibility that that might improve things for you.  You only need the bzroot and bzimage from the distro to test.

 

Well I refreshed everything, moved to version 4.1, reassigned my drives, waited for parity to build and now Im back in business.  I think I either lost one of my drives or created my parity drive on that drive but I can afford to lose the data that was on it.

 

Thanks for all of your help   

originally after replacing my motherboard disk 2 said it had failed.  so in the process of trying to remove the proper drive to ship it back i accidentally ended up hitting the rebuild button (which rebuilds the parity right?)  and now when i boot up the system, the parity drive is orange, disk 1 is red and all other disks are green and say "mounting"  and read value is set to 24.

 

not really sure how to proceed.  i will gladly post my syslog file - but where do i find that?

 

any help would be greatly appreciated.

Just for curiosity sake .. is there a way to get the data off of the drive that I created the parity on?

Just for curiosity sake .. is there a way to get the data off of the drive that I created the parity on?

Sorry to say. No.

 

Parity is not like the files on your other drives... It is instead a series of bit positions, each bit position is used to hold the value of the results of a calculation done on that same bit position on all of your other data drives.

 

So, for a 500 Gig drive, you have 8 bit positions per byte, and 500,000,000,000 bytes.

 

The math to build parity basically looks at the first bit in the first byte on all your data drives and adds up the bits with a value of "1"  If the total is an odd number, the bit on the parity drive is set to a "1"    If the total number of bits with a value of "1" is even across your data drives, the value stored at that bit position on the parity drive is a "0"  That process of looking at bits and calculating what to store on the parity bit is done for each bit in turn on your drives.

 

Calculating parity then examines all the bits on all the data drives, in all bit positions, and does this same calculation writing each parity "bit" in each byte, on each cylinder of the disk based on an odd or even number of bits set to "1" on the data drives.  This effectively overwrites all prior contents on the parity drive.  If you had a partitioned/formatted disk with files before, the partitioning and formatting is gone along with any files previously on the parity drive.

 

This same parity calculation "math" lets us re-create any single missing data disk.    If a disk goes bad, we can look at each bit in turn on the parity and remaining data disks.  If in a given bit position we have an odd number of "1" bits, then the missing bit on the defective drive had to be a "1" since when we wrote parity we set it to a value to have an even number across the bit position.  If in a given bit position we have an even number of "1s" the defective drive must have had a "0" in that bit position, again for an even number of bits set to "1" in that bit position across all the drives.

 

Sorry to say, but although you can use the parity drive to recover a defective data drive, you cannot recover files that used to exist after they were overwritten when a drive was assigned for parity use. 

 

Joe L.