New drives failing to format

[novirium]

Hi,

 

3 times now I've had new drives seemingly fail to format when adding them to the array. Each time the process seems to be the same:

• Stop array
• Add new drive
• Clear new drive
• Clearing finishes - tick the option to confirm I want to format - format new drive
• Array appears to all be working
• As soon as I next stop the array, the new drive lists as "wrong"

 

Once this happens, using "xfs_repair -n /dev/sdc1" indicates the filesystem isn't present:

Phase 1 - find and verify superblock...
bad primary superblock - bad magic number !!!

attempting to find secondary superblock...
.found candidate secondary superblock...
unable to verify superblock, continuing...

(followed by a bunch of repeating dots)

 

Allowing the drive to rebuild seems to work (though obviously takes ages), and after that point I'm successfully able to use the drive in the array.

 

I did note the "Buffer I/O error on dev md4p1" in the log - and I'd suspect something is wrong with the drive, except this exact same sequence has happened with 3 different drives now as I've added them to the array. This time I stopped the array shortly after the format finished just to check if the problem would occur again before I wrote anything else to the drive (and it did). The rebuild process seems to result in working array, but now I'm getting paranoid something else is going wrong and I'm at risk of losing data I put on to it.

 

Any ideas?

 

Thanks

flint-diagnostics-20240116-0955.zip flint-smart-20240116-0958.zip

[JorgeB]

This is a bug, it will be fixed for v6.12.7, it happens if the drive has an existing partition starting on sector 2048, and it did:

 

Jan 16 09:44:53 Flint kernel: mdcmd (5): import 4 sdc 2048 9766435820 0 WDC_WD100EFAX-68LHPN0_JEKXKXVZ

 

For now if you manually wipe the drive (or use UD to do it) and reboot first this will not happen.

 

 

 

[novirium]

Okay, excellent - that sounds like nothing is actually wrong with my hardware then.

 

The array seems to have been fine once it rebuilt the (badly formatted) drive from parity, so I assume this is something that is down at the disk access level, and not a problem present in the emulated drive? As such, allowing it to rebuild the drive repairs the problem introduced by the bug?

 

I'll keep in mind manually wiping any other new drives first.

 

Thanks for the help 👍

[JorgeB]

10 minutes ago, novirium said:

The array seems to have been fine once it rebuilt the (badly formatted) drive from parity, so I assume this is something that is down at the disk access level, and not a problem present in the emulated drive? As such, allowing it to rebuild the drive repairs the problem introduced by the bug?

 

Correct, rebuilding the drive fixes the problem since the emulated disk is not affected.

 

 

[jhughes2466]

This is still happening.  I'm having drives format, lose format and then fail to format habitually.  I can clear them / kill the partition, format, and rebuild.... but its happening at a pretty good rate.  I have 10 disks in array and 4 in the pool and it has happened now across a 60 day span to 4 of the 10 disks in the array.  

 

 

[JonathanM]

14 minutes ago, jhughes2466 said:

This is still happening.

 

On 1/15/2024 at 9:16 PM, novirium said:

3 times now I've had new drives seemingly fail to format when adding them to the array.

Are you the same person? If not, please start your own thread and attach your diagnostics. If this is still you, attach new diagnostics covering the time period when this happens.

 

17 minutes ago, jhughes2466 said:

I can clear them / kill the partition, format, and rebuild..

Rebuild negates all previous actions you listed, the emulated disk includes the format.

[jhughes2466]

1 hour ago, JonathanM said:

 

Are you the same person? If not, please start your own thread and attach your diagnostics. If this is still you, attach new diagnostics covering the time period when this happens.

 

Rebuild negates all previous actions you listed, the emulated disk includes the format.

No, I’m not the same person.  Unless that gets us further towards identifying the solution… in which case then yes I’m the OP’s genetic twin.

 

That said this was the top Google result for this problem, so I wanted to collate together.  Apologies if requirement is to start a new thread, I’ll do that tomorrow morning.  
 

You said a rebuild negates all actions, should I have done that instead?  I noticed after I did the pre-clear and then I lost the data in the drive but the parity sync didn’t seem to restore it.  Where can I find the documentation / guide around restoration in this type of event? 
 

Not new to NAS systems but new to unraid, so apologies again.  
 

-Jeff (the twin)