I have a main cache pool (called cache) and a ZFS pool (called zfs_cache).

The zfs_cache is used by cloud sync service shares only (dropbox and megasync).

 

I just SSHed onto the box and created some folders under my tv and movies shares (using FUSE, so /mnt/user). The shares are configured to use cache and array.

I then moved some content from the zfs_cache to these new folders under /mnt/user/tv and /mnt/user/movies using the commandline (mv).

 

I now have warnings being raised saying my zfs_cache has folders or files from shares that are not configured to use them (i.e. tv and movies).

If I look under zfs_share I can indeed see /tv and /movies, and the new folders I created, but without the content that was moved into them.

The moved content does exist under /mnt/user though, and under /mnt/cache, as I'd expect since new content so will not be picked up by mover yet.

 

If I try and remove the folders (using rm -R) It clears all but the top-level folder rather says `rm: cannot remove 'tv/': Device or resource busy`

 

I'm getting repeatedly spammed by "find common problems".

How do I remove these folders and stop them being created again?

 

I have tried stopping docker and removing.

I have tried stopping the array then restarting and removing.

Nothing works.

 

Please help!

 

Cheers

Edited November 29, 20241 yr by SudoPacman

The original issue was probably a side-effect of you using mv at the Linux level as mentioned in the Caution in this section of the online documentation.

 

Not sure about the remainder though.

Thanks for the response @itimpi, I didn't know that.

It didn't do that in .13, so guessing that's new behaviour?

 

It's a pain though. When I performed the move I was under /user not /zfs_cache, so I don't think it should attempt a rename.

That might explain how it created the folders though. Odd that it didn't then move the files though.

 

I'll have to remember to copy in the future I guess, but I'm not thrilled by the change in behaviour.

 

The question remains, how do I get rid of these folders?

15 minutes ago, SudoPacman said:

t didn't do that in .13, so guessing that's new behaviour?

The caution has applied for years now as it bypasses the Unraid User Share system so not new behaviour per se.   The rest might be.

Just tried rebooting the server and still cannot delete the folders.

Got an unclean shutdown recorded too, so I guess it took too long to shutdown or perhaps there's another bug or change to that, or it's related to this issue?

 

Beginning to wonder whether I should rollback to .13

Edited November 29, 20241 yr by SudoPacman

2 hours ago, SudoPacman said:

If I try and remove the folders (using rm -R) It clears all but the top-level folder rather says `rm: cannot remove 'tv/': Device or resource busy`

This is because the top level folders on zfs are created as datasets, you need to use zfs destroy

 

 

30 minutes ago, JorgeB said:

This is because the top level folders on zfs are created as datasets, you need to use zfs destroy

 

Ahh. Okay. Sorry, I don't know much about ZFS.

I just tried `zfs destroy tv` and it says the dataset does not exist.

 

But after checking the output of `zfs --help` I guessed at:

zfs destroy zfs_cache/tv
zfs destroy zfs_cache/movies

 

And that's done the trick.

Thanks.

 

Still annoying that it does this, but here we are I guess.

 

Thanks for the help.

 

Cheers