JorgeB

With the array stopped assign te new disk as disk3, start array to begin rebuilding. Yes. Only if you have a list of what should be there. Any missing data will not show in the share, it might be in the lost+found folder with a generic name.

Sorry, no, don't remember seeing that error before.

This is not a guide, feel free to post one here, if not I'm going to hide this post since you have already posted the same thing in multiple places.

Cache is always outside the array, you can use the disk share for better performance, map to /mnt/cache inside the server or enable disk shares for SMB.

Old disk appears to be in really bad shape, so for now we can forget about it, disconnect it from the server and rebuild disk3 using the new disk. After that is done you can try using ddrescue on the old disk to see if it can recover some data, but it looks unlikely.

Please post new diags after the mount attempt. I was confusing this with another thread, this disk will likely have some bad sectors and not be readable, at least not completely, so possibly we'll need to consider a different option, but post the diags first.

This is why it's not working: Oct 19 03:19:34 unRAID kernel: mount: attempt to access beyond end of device It might work with a larger replacement device, of course it would require parity be upgraded first.

Disk looks OK, most likely a power/connection problem, check/replace cables/slot to rule that out and as long as the emulated disk is mounting and contents look correct you can rebuild on top.

Safe mode only disables the plugins.

Yes.

There's nothing pointing to a pool problem, try recreating the docker image, you should also update Unraid to latest, you are using a very old release.

You still haven't' posted the diags after array start.

Please post the diagnostics.

@Tuumkeplease don't reply to the FAQ thread, split here.

Depends on how "Enable auto start" is set in Settings -> Disk Settings, but it's not a problem if the array starts, lets to this step by step, start with connecting the old disk3 to the server, go to UD settings and change the XFS UUID for that disk, then you can have the array started and the old disk mounted at the same time, at that point compare the data from the emulated disk3 with old disk3 to see which one is in better shape, note that old disk3 has a SMART problem error that might indicate that data there cannot always be trusted, though in my experience most or even all should be OK, so it would be a good idea to play/check a few files to make sure it looks good.

Filesystem Size Used Avail Use% Mounted on /dev/md1 15T 15T 429G 98% /mnt/disk1 /dev/md1 15T 14T 569G 97% /mnt/disk1 Looks to me like it's a XFS problem with how it's reporting the used space, IIRC there have been a couple of users complaining of something similar recently, if you look at the df output from both diags they agree with the GUI (note that df uses GiB so you need to convert to GB), on the other hand share usage calculation actuality calculates the used space by all the files in a share, it doesn't not look at the free space, so that's why that's correct, my best guess is that you will need to wait for XFS to fix this in an upcoming kernel.

That suggests a hardware issue, how many DIMMs do you have total? If you install just two DIMMs per CPU, using the first 2 slots, does it post?

Easiest way is to mount the old NAS share with the UD plugin then use that path for rsync, it will be /mnt/remotes/ud_mount_point/

Yes, when done you'll need to adjust the services paths if not using /mnt/user.

Run a non correcting parity check, that's a good stress test.

Not easy to find the cause so far but please keep posting the things you are trying to see if we can get to a conclusion, I think it's our best bet for now.

Yes, this changed since v6.10 IIRC.

Nope. You have an unclean shutdown every day?

Use the Dynamix File Manger, easier to move the data without mistakes, then when done do a new config and leave only disk1 assigned.

If you enable the GUI help it shows the supported characters, @ is not supported as part of the user name, it worked before likely due to an earlier Samba version but it won't anymore.