JorgeB

NVMe device was not correctly removed from the pool.
 
Stop array
unassign Samsung SSD from the familia pool
start array
stop array
assign both the Samsung SSD and the Crucial SSD to the familia pool
start array
convert pool to raid1, when done
stop array
unassign the Crucial SSD
start array to remove the Crucial device from the pool
 

Boot using the GUI mode and change the docker network to macvlan, apply, reboot and see if that makes a difference, if it doesn't you can easily downgrade back to v6.11.3 by going to Tools -> Update OS -> Restore

Parity is not yet synced, let it sync first, only after that will turbo write work correctly.

There's filesystem corruption on the cache device, best bet is to backup and reformat.

We usually recommend rebuilding, as long the emulated disk is mounting and contents look correct, other option is doing a new config but you'll need to do a parity check, and that will take as long as the rebuild.

That will work.

Screenshots show the Unraid driver crashing, to me that still points to some hardware issue.

It's being logged as a disk problem for both, and "Elements in grown defect list" not being zero is not a good sign, run a long SMART test for both.

Make sure there's no Android device with a browser window opened on the WebGUI, it can cause issues when the device is sleeping, and to clear the log you can reboot.

You'll need to move those manually. 

You can reboot, parity sync will start over.

Yes, at least one.
 
It still means the RAM is bad, though for servers overclocking is usually not recommended, and XMP is overclocking.

You need to update to Unraid v6.11.3 first, there's a problem with v6.11.2 partitioning >2TB devices, after upgrading repeat the procedure above.

Disk does appear to be failing, xfs_repair cannot repair a disk that returns a read error, hence the input/output error, since there's no parity best bet is to clone it with ddrescue then run xfs_repair.

Possibly there's a container mapped to Data, check all mappings.

Legacy boot is always enabled, EFI or EFI- only allows or not to boot UEFI, if the the USB tool is not working try creating the flash drive manually, but it could be a board/flash drive problem or config issue.

Cache pool is completely full, you must move/delete some data, then better to recreate the docker image.

According to this no shares are being exported by SMB, you need to go into each share that you want exported and enable that.

Docker image is corrupt, delete and re-create.

Cache is completely full, delete/move some data then try again, if it still fails re-create the docker iamge.

Update to v6.11.3 and try again, known issue.

SMART looks OK, but wait the test result, if it passes do a new config with old disk5, re-sync parity, then try the replacement again.

Zeroing the log is IMHO your best bet but since the SSD is also being emulated you can repair that filesystem instead if you want and leave the original device untouched:
 
-start the array in maintenance mode and type
xfs_repair -vL /dev/md4 -when done stop array
-start array in normal mode and the emulated disk4 should now mount, check contents

Set the shares(s) you want moved to cache=yes, disabled the VM service and run the mover.

Upgrade de UD Plugin and reboot.