Can't Start my Array

[IanMcLeish]

My Unraid worked fine after I installed it on version 6.12.4. And I recently upgraded to 6.12.6 a week or so ago, I think. I put everything on ZFS, only because knowing little it seemed to be the way to go?

 

I do have a strange setup because I bought the stuff before I even knew about Unraid, but apart from having to change the USB C cable, which connects the external “DAS” unit to the Protectli computer, because it was causing errors, it has worked without issues for a couple of months.

 

I have a UPS for the Unraid as well, and all was fine until 2 nights ago. In the morning, Jellyfin wasn’t working, and I found I was unable to login to the Unraid GUI remotely, though I was able to ping it. Seeing as it was running in Non GUI mode, all I could do was quick press the power on the computer, and wait. An hour later with no joy, I pressed the button for longer to force a shutdown.

 

Restarting in GUI mode, everything seemed OK, until I attempted to start the array, which seemed to freeze on disk 1 “mounting”. Hours later it was still there, starting the array, and I wasn’t able to cleanly shut it down. I did try shutting down in the terminal, but it wouldn’t so I powered off again.

 

I read a fair bit about pool corruption causing this issue, and I tried to start the array with no pool drives assigned. Fail.

As the Disk 1 was the one which was stuck on mounting, I tried to restart the array with that disk removed. Failed.

I then (Hard shutdown again) and added the disk 1 back. It is now going to be in emulated mode and all data will be destroyed. I guessed with 2 parity drives that might be ok?

 

Memtest ran for nearly 24 hours with no errors. 11 passes, no errors.

 

We did have a bit of a storm the night things went wrong, but nothing else seemed to be troubled the morning I found the issue, and the UPS powers the Disk cage and Protectli computer.

 

I can’t download diagnostics with the array attempting to start, it won’t complete, but I have downloaded 2 system logs, attached as zips and diagnostics with the array turned off. I also attach results of zpool import, I think this clearly shows an issue with disk 1, but why are there 2 disk ones showing in the result??

 

I am not sure if this issue is the same as other threads I have been reading about as mine doesn’t obviously involve cache drives, it seems (to me) the main array that has an issue.

 

I shall stop messing with it until someone PLEASE gives me some advice! I am new to this, so I will need simple instructions please!

 

Many thanks,

 

Ian

screencapture-192-168-0-10-42a43a9b4c02c56ad8f9e8bba99a59f40828e42f-myunraid-net-Main-2023-12-09-21_34_22.pdf unraidnas-diagnostics-20231229-0814.zip unraidnas-syslog-20231228-1101.zip unraidnas-syslog-20231228-1342.zip zpool import result.txt

[IanMcLeish]

Please ignore the screen capture, I uploaded an old one when it was working by accident.

[JorgeB]

Disk1 filesystem is corrupt:

 

status: One or more devices contains corrupted data.
action: The pool cannot be imported due to damaged devices or data.

You will need to re-format and restore from a backup, if one is available.

[Kev600]

Why can't the Parity be used to rebuild this after reformatting? @JorgeB

Isn't that what it's for?

[IanMcLeish]

So I will preclear disk one and put it back in?

[IanMcLeish]

And many thanks for your reply, much appreciated!

[JorgeB]

55 minutes ago, Kev600 said:

Why can't the Parity be used to rebuild this after reformatting? @JorgeB

Isn't that what it's for?

Parity can help with a bad disk, usually not with filesystem corruption, and if I understood correctly, trying to mount the emulated disk has the same result, which is expected.

 

33 minutes ago, IanMcLeish said:

So I will preclear disk one and put it back in?

You can just wipe the disk with wipefs, all data there will be lost, then start the array and reformat.

 

wipefs -a /dev/sdX1

 

Replace X with the correct disk1 identifier, and note the 1 in the end.

[JorgeB]

P.S: it's not normal for a zfs filesystem to get corrupt without a reason, so there may be an underlying hardware issue.

[IanMcLeish]

So I ran that command against disk1 which is listed in array devices as sde, so I ran    wipefs -a /dev/sda1  and this gave some output of deleting blocks, I'm sorry I didn't sopy this output. Maybe I did this the wrong way, but I mounted the sde disk1 using unassigned devices and did the wipe there. Was that the wrong thing to do?

 

Anyway, I put the disk back in as Disk 1 into the array, but attempting to start leaves me in the same place - stuck with Disk1 mounting.

 

I attached another syslog download having left the server running overnight, but things seem to stop at the kernel panic error?

 

Can I ask, even if i do a preclear, will the parity drives rebuild the drive corrupted anyway? Has the corruption been "saved" on the parity drives.

 

I am not sure what is on disk 1, and I realise that I should have a backup of stuff, but I was just setting this up after recovering from multiple failing drives on a Windows Homeserver build from 15years ago. It sat idle for a long time. But I suppose for a simpleton like me, it worked.

 

All the drives were precleared when they were put into the system, which I understand significantly stresses them, and I don't know if it is a good idea to do that again when it was done only 2 months ago, but all these drives are first use, and I know drives can fail early on as well.

 

Thanks for all help suugestions so far!

Ian

 

unraidnas-syslog-20231230-0459.zip

[JorgeB]

Since you didn't post the complete diags I cannot confirm, but I would guess disk1 is disabled and Unraid is trying to mount the emulated disk, which has the same original issue, if that's correct you'd need to do a new config (can check parity is valid) and start the array to format disk1.

[IanMcLeish]

I did a preclear on Disk 1 and a new config. The array is back up and writing parity. I wasn't sure what you meant by "can check parity is valid", but if there was another option to rebuild disk 1 I blew it, but perhaps the corruption would have remained anyway as the array wouldn't start with the drive emulated?

I had VMs on that drive, but only 2 for messing, so no big loss there. I did lose all my dockers, but I copied appdata to my PC, so I'm hoping after the parity is done and mover runs, I'll reinstall the docker images and then hopefully just paste the appdata over the "new ones" - it would take me an age to get them back to the way they were, though SpaceInvader One would get a lot more ad revenue😃

 

Thanks for all your help, it is much appreciated.

[JorgeB]

I believe disk1 was not recoverable in this case.