Please refer to the 6.12.0-rc1 topic for a general overview.
For exclusive shares we made an implementation change. We found issues using bind-mounts with ZFS pools with internal nested child datasets. To overcome this problem, symlinks are created in /mnt/user instead. For example, for an exclusive share named "myshare" which exists only on "mypool" this symlink is generated:
/mnt/user/myshare -> /mnt/mypool/myshare
This implementation is actually a little
Shouldn't be necessary, it's just a troubleshooting step to see if the array stops in a reasonable period of time. There is likely something keeping the array from stopping, causing the shutdown to kill the array prematurely, triggering a parity check on startup.
If that's the case, you need to figure out what is keeping the array from stopping.