Just experienced my worst nightmare.  I recently upgraded a decade old server with new hardware (reused my old hard drives and case).  While assembling the new PC I found an old unused 500gb SSD and decided to include it as a cache drive.  After getting the new system up and running for several weeks, I decided to order another 500gb SSD and mirror the two cache drives.  After watching a couple of SI1  videos, I stopped my running dockers & VMs and disabled both under their respective managers.  I used mover to move all the existing files off the cache to the array.  One file, docker.img, failed to move off the cache, others seemed to incur that same problem, so I manually moved it to the array as they did.  I shutdown and installed the new SSD and assigned it to the pool.  I then formatted the drives as ZFS / mirrored.  I moved the files back to the cache using mover and manually moved docker.img back to the new cache.  I restarted my dockers and everything seemed fine.  Later that evening, I rebooted the server and Unraid reported the dockers failed to start.  Upon investigation it appears the cache pool came up as "Unmountable: Unsupported or No File System", so now I have no dockers and no cache backup.  My initial thoughts are to unplug the cache drives one at a time and restart to see if I can get one of them to mount.  If so, I could at least get the files off for backup.  Diagnostic file is attached.

 

Any help you can give to an old timer would be appreciated.  Please reply assuming you are explaining things to your dog or cat.  My last computer class was Fortran sometime around around 1979 when we used a key punch to poke holes in IBM cards.  Thanks in advance.

 

Currently on Ver 6.12.13, been too scared to upgrade to ver 7.  Probably never upgrade now. lol

tower-diagnostics-20250409-1904.zip

Post the output from:

zpool import

 

That pool appears to be a goner, but that is very odd, both devices having a corrupt label, and should not happen without a reason.

 

Try importing in read only mode, but it will likely fail:

 

zpool import -o readonly=on cache

 

I'm afraid you will need to destroy the pool and create a new one, but I would be worried that it might happen again due to some underlying issue, recommend starting by running memtest.

Thanks JorgeB.  Ran memtest86+ last pm.  4 passes, no errors.  With this being a new build, I would like to find the underlying problem (if any) sooner than later.  Any other thoughts besides ram?

 

Didn't realize there was memtest86 plus and memtest86 plain.  I'm running the plain version now.  Don't really expect different results but don't know anything else to check.

Note that memtest is only definitive if it finds errors, but could also be a CPU or controller issue, for example, or something was manually done to the devices, or even a zfs bug, though the latter seems the most unlikely to me, but not impossible.

Just to wrap this up.....

Ran memtest86, memtest86+, and OCCT for several hrs each w/o finding any errors.  Remounted the two SSD's as a cache pool and couldn't format them as ZFS?  Auto was the only option and it was greyed out.  Formatted the drives (turned out to be btrfs) with everything appearing to work normally.  Deleted the cache pool and unassigned the cache drives.  I then recreated the cache pool again using the same two SSDs.  This time it allowed me to choose ZFS for the file system and the drives formatted OK.  I've been using the new cache pool for several days restoring dockers and reorganizing my Plex layout.  Everything is working incl Mover moving files from cache to the array so I don't have a clue what happened initially.

The only thing I'd like to fix before backing everything up is my Sonarr path.  I cannot get my root folder set to save my butt.  Not sure what the problem is.